String replacement preparation for file comparison

[TJRandall]

I need to compare CVS files to Oracle code extracts from our Production environment. The issue is that some of the CVS files do not declare the function (package, procedure) in the same format as the Oracle extract, which later flags the files as different when using Text:Diff.

This difference:

CREATE OR REPLACE FUNCTION DC_F_COD_RANDOM_MMA RETURN VARCHAR2 IS

vs.

CREATE OR REPLACE FUNCTION "TRON2000"."DC_F_COD_RANDOM_MMA"
         RETURN VARCHAR2 IS

is one that I can safely ignore.

My approach is to open each CVS file, and check to see if it declares in the Oracle format. If not, replace the current declaration with a cleaned up Oracle one. (this is actually my first question - does this sound like a valid approach?)

So I look for the piece between 'CREATE OR REPLACE' and 'AS' or 'IS'. Inside there, I will find 'FUNCTION', 'PROCEDURE' or 'PACKAGE', and the actual name. So in the code I'm working on below, I have a function which is not in the Oracle format - so I have to change the file contents to

CREATE OR REPLACE FUNCTION "TRON2000"."DC_F_COD_RANDOM_MMA"
         RETURN VARCHAR2 IS

Is there a way to do this cleanly / easily, without all the IF/ELSE blocks that I'm building out - my current approach feels like I'm going to be sprawling quickly into something that is unmanageable.

Your thoughts / pointers / advice is GREATLY appreciated!

#!/usr/bin/perl -w

use strict;
use warnings;
use 5.010;
$/=undef;   
my $start_search_str = 'CREATE OR REPLACE ';

while (<DATA>) {
   my $rc;
   
    # if the file contents are in the Oracle extract form
    if( $rc = /$start_search_str (.+) \"TRON2000\"\.\"/ .. /(AS|IS)/  
+){
      # file is the right format - get the next one
      last;
    } else {
      # nope - need to fix them - first - get the part of the string 
      # between 'CREATE OR REPLACE ' and (AS|IS), then insert
      # "TRON2000"."<name>" around the name
      
      my $contents = $_;
      $contents =~ /(($start_search_str)(.*)(AS|IS))/si;

      my $function_name = $3;
      if ($3 =~ /FUNCTION/si) {
        $function_name =~ /( FUNCTION(.*) RETURN)/si;

        print "NOW: function_name: $function_name\n";
        
        # now substitute and write it back to $rc
      };
      
    };
}

__DATA__
CREATE OR REPLACE FUNCTION DC_F_COD_RANDOM_MMA
         RETURN VARCHAR2 IS
-- 
  l_random_code1               VARCHAR2(1) ;
  RETURN l_random_code ;    
--       
END dc_f_cod_random_mma ;

[roboticus]

TJRandall:

I recently had a problem where someone *radically* reformatted a package, and I had to add one of their patches to the non-screwed-up version. So to find the differences, I reformatted *both* versions and diffed the results. Then I could identify *what* the changes were, so I went back to my original version and applied the appropriate patch.

The reformatter was trivial: I eliminated all line breaks, then I inserted line breaks after each semicolon and comma, like so:

$ cat squidge.pl
#!/usr/bin/perl
use strict;
use warnings;
use File::Slurp;

my $INFName = shift or die "Missing file name";
my $t = read_file($INFName);

$t =~ tr/\n\r/  /;
$t =~ s/;/;\n/g;
$t =~ s/,/,\n/g;

open my $FH, '>', $INFName . ".out" or die;
print $FH $t, "\n";

Then I used GNU diff, telling it to ignore whitespace differences and such:

diff -abiwEB --strip-trailing-cr file1 file2

If you're just trying to identify files that differ, this might do the trick for you.

...roboticus

When your only tool is a hammer, all problems look like your thumb.

[TJRandall]

This is a great approach - thank you! This fixes my "other" compare problem (differences flagged in Text::Diff on carriage returns). I still have to figure out the first "fix" - taking the function/package/procedure name and making sure it is:

CREATE OR REPLACE FUNCTION "TRON2000"."<name>" AS.

Thank you again!