I've written some tools which will take a bunch of .lumi files from an AC++ job
and tell you how much luminosity was processed according to a good run list.

The tools are located in /cdf/data10a/ucntUtils/lumi and are meant to be run 
with the development version of the cdf software. 
****
****EDIT 07 Nov 2007 YOU NEED TO USE CDFSOFT2 6.1.4 FOR LUMSUM.PL
****
The first tool is sort_lumi.tcsh. It takes a directory containing .lumi files 
from an AC++ job, creates a lumi/ subdirectory in the directory, and puts .lumi
files parsed by run number. 

For example, the directory /cdf/data10a/Datastes/UCNT_533nt_59719x/DATA/ELECTRON/bhel0d/01/
has a number of .lumi files, each containing many run numbers. Executing

sort_lumi.tcsh /cdf/data10a/Datastes/UCNT_533nt_59719x/DATA/ELECTRON/bhel0d/01/

will produce the directory

/cdf/data10a/Datastes/UCNT_533nt_59719x/DATA/ELECTRON/bhel0d/01/lumi/

with files like bd021cb9.0001hel0.lumi.138425 which contain the number of 
luminosity information in file bd021cb9.0001hel0 for run 138425.

This should be done once for each directory.

The second tool is sum_lum. This takes as arguments a good run list and a list 
of directories, and returns the run numbers in the directories and the 
offline luminosity corresponding to those run numbers, as well as the sum of 
all of the runs. All luminosity numbers are in pb^-1. If you want to use
all runs in the datafile, there is a list of all runs in the DFC found in 
allruns.list as well as a python script allruns.py which can be used to update 
this list. So,

sum_lum -grfile allruns.list -i /cdf/data10a/Datastes/UCNT_533nt_59719x/DATA/ELECTRON/bhel0d/01
****
****EDIT 07 Nov 2007 FOR GR LISTS WITH RUNSECTIONS USE SUM_LUM_RUNSEC
****

will produce the following output:

...
...
...
179096 0.065
179097 0.057
179099 0.026
--------------------------
TOTAL:	55.576 pb^-1

Note, the directory you give it is presumably the one containing your ntuples 
and lumi/ as a subdirectory, not the lumi/ directory itself. 

You can give it multiple directories, and it will print out the sum of all of 
the luminosities. However, it does not handle overlaps of runs and run sections
smartly and will simply double count them. Work on this can be done in the 
future, but this is a start. For single datasets (like bhel0d or bhmu0d), there
shouldn't be any overlaps. However, there can be much overlap between two 
datasets.

If you have any questions, email me (cwolfe@hep).

					-- Collin