Shell Programming and Scripting

At work, we use a software development product (from a company that will remain nameless, but whose name may be considered a synonym for "logical"). The development trees are organized beneath a top directory, let's call it "$rat". The first level under $rat contains the major system names, and at that level there is either the list of subsystems or one additional directory layer then subsystems. I.e., using .1 for subsystems,

Code:

$rat
   skeletal
      protective.1
      support.1
      ambulatory.1
   circulatory
      blood
         redcells.1
         whitecells.1
         plasma.1
      endocrine.1
   dermal
      withhair.1
      withouthair.1
   muscular
      ambulatory.1
      manipulative.1
   sensory
      visual
         red.1
         green.1
         yellow.1
      olfactory.1
      ...

Beneath each subsystem, each there are copies of the views for each developer interested in developing there as well as the baselined views of the software. I.e., using .2 for the views,

Code:

$rat
   skeletal
      protective.1
         fred.2
         barney.2
         wilma.2
         baseline.2
      support.1
         fred.2
         barney.2
         betty.2
         baseline.2
         ...

So fred, barney and wilma do protective skeletal development, while fred, barney and betty do support skeletal development.

Here's the question: Is there a way to write a single "find" expression that will find, for every system, all the source code in betty's view, or (if betty doesn't have a view on the given code) all the source in the baseline view?

The best I've been able to come up with is a two level find:

Code:

find $rat -name '*.1' -prune -exec findbetty '{}' |
  while read viewname ; do 
    find $viewname -name '*.[ch]'
  done > output.txt

There are a few more clauses in both find commands to reject "logical's" tool-maintaintained directories.

findbetty is a script that does this:

Code:

#!/bin/sh
ls -a $1 |
  awk -v base=$1 '/betty\.2/    { print base "/" $0 ; found = 1 }
                  /baseline\.2/ { d = base "/" $0; }
                  END { if (!found && length(d)) print d }'

I.e., if in the given subsystem ".1" directory, there is a "betty.2", print it, otherwise if there is a "baseline.2", print that.

I'd like to be able to do this in a single "find" command with no calls to external programs.

Ideas?

For more complex shell scripts, I suggest you use either PHP or PERL. You will have much more overall flexibility and capability. After using these more advanced shell, I rarely do anything in older shells anymore, except very simple scripts (and even the more simple scripts are fun in PHP and PERL).

Of interest, maybe:

www.hotscripts.com

Six months ago, I would have written this in Perl or Ruby. But if you look at the part that does the work, it's only 4 lines long --- 8 if you count the subservient ls/awk script --- which doesn't qualify as "more complex" in my mind. I guess I could use find2perl to prototype the Perl expressions I'm interested in, though ISTR you take about a 20% performance hit by going through Perl (not a big deal here).

FWIW, I save the output from this program, then from vi I execute another script that picks out the source file I want and prints its full path, as in

Code:

:e `findfile zzz.c`

It took me a while to understand what the vi intro doc meant when it said that arguments to ":e" are passed to the shell for expansion before trying to edit the file. But I think I've got it now ...