I wrote macro that works such way.
Freely available from //http://en.feautec.pp.ru/SiteNews/dofilelsp
Comments are welcome.
Thanx to Lutz for notes about programming :-)
Very nice! Another possibility would be to use 'parse' to break up the file at the delimiter and then process the records, i.e:
(dolist (rec (parse (read-file "/etc/passwd") "n"))
(println ">" rec))
wrap you macro logic around it to isolate the loopvariable, filename and string delimiter and perhaps it comes out much shorter.
Lutz
Yes, I know about (parse (read-file)). It pretty small itself to be sufficient without (dofile).
But if the file is many-megabytes sendmail log, sales report for a year or so, then reading whole file will not be memory efficient... Moreover, we usually need only few fields from each record... And usually a few records from whole file.
I trying to write unix-like filter, that will conform to lispish processing style.
Perhaps syntax something like awk.
Then rather than do-file the command line would feed multiple files/file names in. viz:
(BEGIN (setq total 0))
( (regex "lab+") (inc total))
(END (println total))
where regex with 1 parameter matches against the automatic readline.
Nigel
Quote from: "nigelbrown"
Dmi,
I see you already have awklsp at http://en.feautec.pp.ru/SiteNews/awklsp
Nigel
Yes. I like awk :-)
But my exersices show that awk.lsp gives more awkish, than lispish and while coding in newlisp I prefer to go slightly different way.
Nevertheless (rawk) in awk.sp is either good, I think :-)