From: Florian Gilcher Date: 2012-10-23T07:00:32+09:00 Subject: Re: Models classes from CSVs On Oct 22, 2012, at 11:33 PM, Todd Sanders wrote: > I have 8 csv dumps that I receive nightly. I have to transform them > into XML and FTP them. So far I have been able to complete this task. > Now the requirements are to match some records up between csv dumps. > > I did not want to load all the files into a database, as I only work > with each nights data. I thought I could convert the csv to a model > after parsing it. I am unable to find information on converting the CSV > to active record or an active-model (not sure either one of these ideas > are correct approach). ActiveRecord requires a database. Persisting data in an SQL-database is the whole point of the library. ActiveModel is an interface that generalises common tasks with model-like data. It works by giving you a general implementation in which you have to fill out the blanks. ActiveRecord one of such implementations. Writing a fully-fledged AM-implementation for a bit of CSV data might be a bit too much. There are multiple approaches: 1) Just use plain old ruby objects and than match them by arranging the corresponding objects in memory. Its a bit more legwork, but depending on your dataset size, it shouldn't be too difficult. 2) If the task is more complex, why not connect ActiveRecord, DataMapper or Sequel to an in-memory-SQLite database ("sqlite://:memory:", afaik)? This gives you all the power of the corresponding lib, but without the headache of handling a DB server. I hope this gives you a few pointers. Without knowing the actual tasks and the size of those files, its hard to make good suggestions. Regards, Florian