Re: Multi-line record headers

In article
<83035e9f-37b2-4264-9df0-a1359a007c96@[EMAIL PROTECTED]
>,
 <kevin.byford@[EMAIL PROTECTED]
> wrote:
% This one has me stumped.  In the below example I'm trying to get the
% (customer) number (which might also contain capital letters) above
% each group of contacts to show on every line in front of each
% contact.  There are no tabs in the file, just spaces.  Some customer
% numbers have zero contacts, some have many.

The key to solving this kind of problem is to find a reliable way
to distinguish between the different kinds of data in your input.

% Here's the source:
% 
% 00270
%  OE DENNIS ANSZ                   OWNER/EXECUTIVE      600/200-0110
% 
% 00275
%  OE JAMES ARNILD                                       310/755-2079
%  ZA SHARI                         ACCOUNTS PAYABLE     310/755-2079

From this, it seems that the contract information can be identfied
by the leading white-space, and the customer number can be identified
by the leading not-white-space. So, for a start:

 /^[^ ]/ { customer = $0 }
 /^ / { print customer, $0 }
 !NF # this is to preserve blank lines

The trouble with this is that it throws out the contractless customers.
I'd handle that by setting a flag.

 /^[^ ]/ { customer = $0; needsprinting = 1 }
 /^ / { print customer, $0; needsprinting = 0 }
 !NF && needsprinting { print customer }
 END { if (needsprinting} print customer }
 !NF # this is to preserve blank lines

-- 

Patrick TJ McPhee
North York  Canada
ptjm@[EMAIL PROTECTED]