From: Albert Wagner <alwagner@...>
Date: 2001-10-15T09:13:10+09:00
Subject: [ruby-talk:22539] Re: extracting from delimited text files

Cool!  That really works.  Thanks, Bill.  
All I have available at the moment is the pickaxe book.  The only mention of 
a use for  "?" in Chpt 5 is as a repetition modifier.   I just found the 
extensions in Chpt 18.  But, seriously, I don't think I could have figured 
out a use for them without an example.  Thanks again.

On Sunday 14 October 2001 06:17 pm, you wrote:
> Hi Albert,
>
> From: "Albert Wagner" <alwagner@tcac.net>
>
> > Thank you, Robert.  I was unaware of both the csv module and the Ruby
> > Cookbook.  I got this far with String#split but couldn't get rid of the
> > quote in element 0 and how to force empty strings for the trailing
> > commas:
> >
> > irb(main):048:0> x
> > "\"aaa\",,\"ccc\",,,\"fff\",,,"
> > irb(main):049:0> y = x.split(/\"\s*,\s*\"|\"\s*,|\s*,\s*\"|\s*,|\s*\z/)
> > ["\"aaa", "", "ccc", "", "", "fff"
>
> You can keep the trailing empty fields by passing a 'negative limit'
> into split, e.g. y = x.split(/regexp/, -1).  However, the elimination
> of the leading quote on the first field seemed like a stumper (without
> producing an unwanted initial empty field.)  Here's a regexp solution
> using String#scan . . .
>
> require 'runit/testcase'
> require 'runit/cui/testrunner'
>
> def csv_split(str)
>    
> str.scan(/(?:\A|,)\s*"([^"]*)"|(?:\A|,)([^",]*)(?=,|\z)/).flatten!.compact!
> end
>
> class TestCsvSplit < RUNIT::TestCase
>     def testCsvSplit
>         fields = csv_split(%Q{"aaa",,"c,d,",,,"fff",,,})
>         assert fields == ["aaa", "", "c,d,", "", "", "fff", "", "", ""]
>     end
> end
>
> RUNIT::CUI::TestRunner.run(TestCsvSplit.suite)
>
>
> Hope this covers all the cases.  :-)  BTW this doesn't handle
> "embedded quotes" within a quoted string.  That is, the example
> you provided didn't specify how or whether embedded quotes
> within quoted strings in the CSV fields were treated.  This
> regexp (above) won't deal with quotes [escaped somehow] inside
> a quoted string.
>
> Hope this helps,
>
> Bill