Back when I worked for Miltec, we were doing using huge (10-20mb) space separated data files to drive visualizations. The data coming from various sensors would vary in the order that it recorded yaw, pitch and roll and since the data files had no headers, it worked best to simply run them through a preprocessor to put them in a uniform order. This is that program. I got fancy and added some options. I don't really remember how to run this program. Here's the help though:
restructures a ascii text file which is delineated into columns
separated by spaces or columns.
-r "[colnum[:tranform]]+" -> restructure
colnum = column number for the output file
transform = factor to transform the output by
exe: -r "3:10 4 5 1:.01 3" would generate an output with the 3rd, 4th,
5th, 1st and 3rd columns of the input with each element of the first
3rd column multiplied by 10 and each element of the original 1st
column (the 4th in the output) multiplied by .01.
-p -> pad
exe: -p would generate output columns padded to the length of the
longest element in the column
-d "separator" -> output column separator; default: " "
-i "filename" -> input filename
-s "rexex" -> input separator regex; default: ",\s+"
-x -> xmlize
exe: -> -x would generate output of the form:
<data name="dataname">
<column name="colname">
<point index="1">123456</point>
<point index="2">123456</point>
</column>
<column name="colname">
</column>
</data>
-n "dataname" -> title for the data
exe: -n "2001Earnings" would put the string 2001Earnings as the name
property for the xml output or would use the table 2001Earnings as
the table to insert into in the database.
-b "[user][:pass]@[hostname][/dbname]" -> insert columns into database url
exe: will:mypass@localhost/samples will insert the columns into the
table specified by -n; the table must already be created and have
column names matching the columns of the input.
-h -> help