CSVHelper: An Objective-C repository from halmueller

This project generates Objective-C code to support the conversion of
CSV files into Core Data. 

The heart of the project is Matt Gallagher's CSVParser class. This
class reads a CSV file and calls a designated @selector for each row
of the CSV, passing the selector an NSDictionary of values from that
row, keyed by the column headers. I've used this class in several
projects and it works quite well.

For my current work, I'd like to read about 2 dozen files, each with a
different layout. Manually creating an Objective-C class for each of
these files, and writing the associated calls and callbacks, felt like
too much work. I automated it instead.

The resulting class is CSVDirectoryProcessor. A CSVDirectoryProcessor
reads every CSV file in a directory. It then does the following:
 — creates Foo.h and Foo.m class stubs for each Foo.CSV found
 — generates a partial MyDocument.m (containing all the callbacks)
which reads each CSV and then invokes CSVParser on that file. Add this
to your MyDocument.m, editing the assumed self.managedObjectContext if
needed. 
 — generates a summary.txt which lists each class created, and the
properties in each class. This summary.txt is useful as a reference
for manually editing your .xcdatamodel file.

CSV files are assumed to have a header row. That's how the properties
of the generated class are named.

Code generation is different depending on whether you select vanilla
Core Data, Core Data plus MOGenerator, or simply NSObject
subclasses. If you select MOGenerator, you still need to invoke
mogenerator to generate the _Foo.h and _Foo.m files from your data model.

The supporting project, CSVHelper, is just a GUI wrapped around
CSVDirectoryProcessor. It lets you navigate to a directory, choose
what sort of code generation you want, and fire off the code
generator. All generated source code is placed into the same directory
as the original CSV files.

Once CSVDirectoryProcessor has run, you still need to edit the
.xcdatamodel file, create all of your entities, and wire up the
relationships. The assignment of values into Core Data
attributes/properties is quite crude, and you'll certainly want to
edit the generated callbacks. But this project gets a good portion of
the tedium out of the way.

Matt Gallagher's original CSVParser is described in this blog post:
http://cocoawithlove.com/2009/11/writing-parser-using-nsscanner-csv.html

MOGenerator is available at https://github.com/rentzsch/mogenerator.

Finally, the many CSV files I'm working with were very nicely
generated by Jakob Egger's MDB Viewer, a Mac OS X tool that reads
Microsoft Access files and creates CSV or SQL. It's on the Mac App
Store at http://itunes.apple.com/us/app/mdb-viewer/id417392270?mt=12.

The "samples" folder shows three different variations of code
generated from Matt's Australian Postcodes sample (which I modified,
to add header rows).

Hal
halmueller/CSVHelper
