Searching for a sequence in a text

fixed the issue. Will look into that further Why not search for the pair you want? Also, you can use \s to match whitespace.

import re

def find_pair(x, y, mystring):
    return re.findall(r'\(\s*?%d,\s*?%d\s*?\)+' % (x, y), mystring);

print find_pair(2, 4, "(3,4) , (2,   4),(5,4), (2,3), ")

may help you . Overall, searching for numeric equality will probably be faster than a string comparison, helped also because the size of the documents searched will be smaller. It's hard to say though whether the speed increase will be significant, since the performance will be impacted by a number of factors that might dwarf a string vs integer comparison - for example how the size increase of the documents affects what's in RAM, or whether doing another query for the string each request is slower.
Indexes will be much faster.

hope this fix your issue If you filter the log to only show lines that match "ERROR" or "Import #" then you can use the index of the ERROR line to retrieve the previous line. This line should be the import line as we are filtering the log only to show imports and errors.
This is making an assumption based on the log you have sent. It is possible this may not work if the log contains other lines that match these patterns that are not relevant to our lookup.

# Get content from log file where line matches 'ERROR' or 'Import #'

$log = gc c:\log.txt | ? { $_ -cmatch 'Import #' -or $_ -cmatch 'ERROR' }

# For each line in log file

foreach($l in $log)
{
    # If it matches ERROR

    if($l -cmatch 'ERROR')
    {
        # Return the previous line

        $log[$log.IndexOf($l)-1]
    }
}

around this issue If Perl is acceptable for you, it will be easy to do what you want because Perl is good at handling mult-line regular expressions. Then how about:

perl -e '
while (<>) {
    $text .= $_;
}
$add = "\nAFFECTED_ROWS := AFFECTED_ROWS + SQL%ROWCOUNT;";

$text =~ s/^UPDATE[\s\S]+?SET[\s\S]+?;/$&$add/mg;
$text =~ s/^DELETE[\s\S]+?FROM[\s\S]+?;/$&$add/mg;
$text =~ s/^INSERT[\s\S]+?INTO[\s\S]+?;/$&$add/mg;
$text =~ s/^MERGE[\s\S]+?INTO[\s\S]+?(WHEN MATCHED THEN|WHEN NOT MATCHED)[\s\S]+?(UPDATE|INSERT|DELETE)[\s\S]+?;/$&$add/mg;

print $text;
' inputfile