[SOLVED] Help in script - Getting table name from a text file

[wiss]

Hi guys,

I have been working on this since 3 days but wasn't able to achieve what I want

I have a big text that has the following format:

Current max fieldLen for table1 (a):
Fld# Width MaxWidth ERR NAME
---- ------ -------------- ------ ---------
2: 80 38 *** field-name
3: 4 2 field-version
4: 40 7 field-value
5: 90 116 value-meaning

Current max fieldLen for table2 (b):
Fld# Width MaxWidth ERR NAME
---- ------ -------------- ------ ---------
2: 80 38 field-name
3: 4 2 field-version
4: 1998 720 definition
5: 2 2 data-type
8: 40 12 field-format
10: 4 2 data
11: 80 19 split-length

Current max fieldLen for table3 (c):
Fld# Width MaxWidth ERR NAME
---- ------ -------------- ------ ---------
2: 10 5 uic
3: 2 1 user-type
4: 4 2 chess-rels
5: 4 0 next-rels
6: 6 0 next-rels-active
7: 16 3 *** security-format
8: 30 6 default-applid
9: 15 6 next-txid
10: 80 19 user-desc
12: 16 5 ci-acc-group
13: 40 0 ci-password

Current max fieldLen for table4 (d):
Fld# Width MaxWidth ERR NAME
---- ------ -------------- ------ ---------
3: 40 9 type
2: 120 193 *** description

Current max fieldLen for table5 (e):
Fld# Width MaxWidth ERR NAME
---- ------ -------------- ------ ---------
2: 10 5 aic
3: 16 5 name
4: 30 8 austpac-address
5: 10 0 user
26: 20 5 route-profile

As you see, the column headers are fix. Also, I have the following fixed words in the beginning of every table "Current max fieldLen for".
For some of these lines, in the error column, there are 3 stars.

Note: The table name is the word that could be found after "Current max fieldLen for".

Basically, I am searching for a way in order to get the table name in case the table has "***" in the error column, in any of its fields.

For example, for this sample I provided, I need the output to be exactly:

table1
table3
table4


Why not table2 or table5? Because they don't have any "***"

For now, I did a little bit of progress by using:
"
egrep "(^Current|\*\*\*)" filename.txt > test.txt
cat test.txt | awk '{print $5}' > test2.txt
"
Could anyone help please?

Very much appreciated!

Regards.

NOTE: Sorry for the bad format of the text file. I am attaching an image with the proper format

[Kloschüssel]

1] That image is not of a great help
2] Print table names:

#!/usr/bin/awk -f
/Current max fieldLen/ {
print $5
}

3] The tricky part is extending the above into a script that prints only table names with ***, so let's try to output only ***:

#!/usr/bin/awk -f
/:/ {
if ($4=="***")
print $4
}

4] And finally combine it into the final script:

#!/usr/bin/awk -f
/Current max fieldLen/ {
table=$5
}
/:/ {
if ($4=="***")
print table " has " $4
}

Testing it produces the following output:

$ ./test.awk test.txt
table1 has ***
table3 has ***
table4 has ***

I'm a awk novice and all i had to read was written here: AWK Programming

--- EDIT

#!/usr/bin/awk -f
/Current max fieldLen/ {
        table++
        result[table,0]=$5
        result[table,1]="false"
        result[table,2]=0
}
/^[0-9]*:/ {
        result[table,2]++;
        if ($4=="***") {
                result[table,1]="true"
        }
}
END {
        print "table    |       ***     |       fields"
        print "---------------------------------------"
        for(i=1; i<=table; i++)
                print result[i,0] "     |       " result[i,1] " |       " result[i,2]
}

[wiss]

Thanks a lot for your help!!

I will try it soon and let you know the outcome. Sorry about the photo, i didn't realize it was so small!!!

[drl]

Hi.

Using your data on file "data1" with perl script "p1":

#!/usr/bin/env perl

# @(#) p1	Demonstrate reading in "paragraph" mode.

use strict;
use warnings;

my ($table);

$/ = "";    # Read "paragraphs" as a single line

# Table name is string after "Current max fieldLen for"

while (<>) {
  m{Current max fieldLen for\s+(\w+)};
  if ( defined($1) ) {
    $table = $1;
    print "$table\n" if m{[*]{3}};
  }
}

exit(0);

produces:

% ./p1 data1 
table1
table3
table4

I used the property of your data that "***" does not appear elsewhere in a table.

Best wishes ... cheers, drl

[wiss]

Hi Kloschüssel,

I am not sure why this script is not going through the whole table. It's just displaying the first 2 tables that have "***"

[wiss]

Hi DRL,


Thanks a lot for your help! It looks working perfectly

In case I have tables that have a dash included in the name, this script just picks the first part. example:
user-id will be displayed as user

Any idea how to take the whole name? (ie. user-id)

Thanks in advance

[drl]

Hi.

Adding to your dataset with a name such as you described, and a changed perl script line:

m{Current max fieldLen for\s+(\S+)};

I get:

% ./p2 data1 
table1
table3
table4
name-with-dashes

Best wishes ... cheers, drl

[Kloschüssel]

I am not sure why this script is not going through the whole table. It's just displaying the first 2 tables that have "***"

Can't tell without an example. Probably those lines do not match the regex '/Current max fieldLen/' or '/^[0-9]*:/'?