Update: Since you want to flag the last record for each patient/record-type (if I read that correctly), you will have to decide whether you want to keep records in memory (if there aren't too many records or distinct patient/record-types), or load the data into a database (even a lightweight database like DB_File or DBD::SQLite might be appropriate) and sort by patient/record-type/record-number(desc) to easily "flag" each "last" record. But know that (as already mentioned) perl is perfect for the job, and it would be best if you would start writing sample code and ask for help with specific problem areas (rather than me guessing at what you might have problems with, and which approach to take :-).

I do this nearly every day at work. I'm always chopping up files and creating new files in order to figure out something interesting from the data. Perl is a great tool for this.

You'll want to use the split function for delimited files; unpack or Parse-FixedLength, etc., as runrig suggested; and regular expressions (regexes) for parsing human-readable reports. Actually, I frequently use combinations of these techniques in the same file.

One suggestion: I've found it handy to write some simple domain-specific tools (such as a transaction dumper...) that accept some simple data format or can make some assumptions about the structure of the data. Then you can write quickie perl scripts to extract the appropriate data from your dataset to dump with your tools. I have lots of little chunks of code that do such tasks.

As an example, in my job, I frequently have to find merchant numbers in one file and match them with merchant numbers from another file. Our merchant numbers are always in 16-character fields, so I have a little program that accepts four parameters (Filename 1, MID column, Filename 2, MID column). It simply scans the first file and collects merchant numbers, then scans the second file and prints out hits. I'm not at work right now, but it goes something like this:

#!/usr/bin/perl -w
my $usage="
merch_match <FName1> <MID Col> <FName2> <MID Col>

Prints all Merchant IDs found in File 1 that are also
contained in File 2.
";

my %MIDS;
my $FName1 = shift or die $usage . "Missing FName1";
my $MIDcol1 = shift or die $usage . "Missing MID COL #1";
my $FName2 = shift or die $usage . "Missing FName1";
my $MIDcol2 = shift or die $usage . "Missing MID COL #2";

# substr uses 0-based string offsets
$MIDcol1--;
$MIDcol2--;

# Gather MID numbers from File 1
open(INF,'<',$FName1) or die $usage . "Can't open " . $FName1;
while (<INF>) {
    my $M = substr($_, $MIDcol1, 16);
    $MIDS{$M}=0;
}
close(INF);

# Gather MIDs from File 2
open(INF,'<',$FName2) or die $usage . "Can't open " . $FName2;
while (<INF>) {
    my $M = substr($_, $MIDcol2, 16);
    ++$MIDS{$M} if exists $MIDS{$M};
}
close(INF);

# Print match list
for my $M (sort keys %MIDS) {
    print $M if exists $MIDS{$M};
}
[download]

--roboticus

Depending on what you want to do with the data, these manpages that come with the perl distribution might be useful: open, split, pack.

This is a simple example to get you started:

$/ = \ 7 * 1024; # read 7 K records.

while (<>) {  # read record into $_
   my $info = substr($_, 0, 1000); # put first 1000 bytes in $info
   my $lab = substr($_, 1000, 1000);
   my $diagnosis = substr($_, 20000, 49999);
   my $last = chop; # get last character

   # now do stuff with that data

}
[download]

If there are a lot of fields in the record (which admittedly has not yet been established), substr is what I definitely would not use (*yuck*).

Yes, Perl would be perfect for the job.

/J\

As for your actual question, I agree that Perl sounds like it should work pretty well for you.

Though I admit that database has a structure far more complex than anything I have dealt with in Perl, I think it would be a fine language to use, assuming your system has enough memory (even then, you can just process one or a group of lines at a time).

_________________________________________________________________________________

I like computer programming because it's like Legos for the mind.