First match if we are on the first line of input, if so print the whole line, else print the first word followed by a comma

Question

for eg:

CREATE TABLE MWWDATA.ACK997 (
    AKTYPE CHAR(2) DEFAULT ''  NOT NULL ,
    AKNUM CHAR(9) DEFAULT ''  NOT NULL );

CREATE TABLE MWWDATA.APREIDEXC (
    EMPLID NUMBER(15, 0) DEFAULT NULL );

I want output like:

CREATE TABLE MWWDATA.ACK997(AKTYPE,ANUM);
CREATE TABLE MWWDATA.APREIDEXC(EMPLID);

Answer 1

The sed way

As sed is a lot lighter than mosts alternatives, there is another solution

sed -ne '
    /CREATE/{h;d};
    s/^ *\([^ ]\+\) .*\()\);\? *$/\1\2/;
    ta;
    s/^ *\([^ ]\+\) .*$/\1/;
    H;
    bb;
   :a;
    H;
    x;
    s/\n//;
    s/\n/,/g;
    s/ (/(/;p ;
   :b'

This could be written:

sed -ne '/CREATE/{h;d};s/^ *\([^ ]\+\) .*\()\);\? *$/\1\2/;ta;
    s/^ *\([^ ]\+\) .*$/\1/;H;bb;:a;H;x;s/\n//;s/\n/,/g;s/ (/(/;p ;:b'

This will render, from you sample

CREATE TABLE MWWDATA.ACK997(AKTYPE,AKNUM)
CREATE TABLE MWWDATA.APREIDEXC(EMPLID)

Answer 2

This perl script

• prints the full line (without a \n) if it matches /^CREATE TABLE/
• the first word followed by ); and a newline if the line ends with ;
• and the first word followed by a , if the line ends with a ,.
• All other input lines are ignored.

#! /usr/bin/perl

use strict;

while(<>) {
   chomp;
   if (m/^CREATE TABLE/) {
       print $_;
   } else {
        my @line=split ;
        if (/;\s*$/) {
            print "$line[0]);\n";
        } elsif (m/,\s*$/) {
            print "$line[0],";
        }
   }
}

The script's output matches what you asked for:

$ ./summarise.pl table.sql 
CREATE TABLE MWWDATA.ACK997 (AKTYPE,AKNUM);
CREATE TABLE MWWDATA.APREIDEXC (EMPLID);