From: "danbernier+ruby@..." <danbernier@...>
Date: 2006-07-08T05:00:11+09:00
Subject: Gen. Row objects from tab-delim text lines

This is nothing ground-breaking, but I thought I'd share it anyway.  It
was a fun little exercise, and I've found it pretty useful.

RowParser.parse takes a String (or will read in an IO), assumes it's
\n- and \t-delimited, and that the first line is \t-delimited column
headers.  It generates a Row class with attributes named for the
columns, and returns an array of populated Row objects.  It does a bit
of name munging, and allows for empty columns.

It's been useful during this data conversion I'm working on -- both SQL
output and MS Word/Excel tables save to tab-delim text files easily.

Example:  here's a table of nutritional info for different beers (from
http://www.realbeer.com/edu/health/calories.php).  I hope the tabs come
out alright -- notice that some of the columns are empty:

Brewery/Brand	Beer	Alcohol %	Calories	Carbs
Amstel Light	Amstel Light	3.5	95	5
Alaskan Brewing	Alaskan Amber	5
Alaskan Brewing	Alaskan Pale Ale	4.6
Alaskan Brewing	Alaskan Stout	5.7
Alaskan Brewing	Alaskan ESB	5
Alaskan Brewing	Alaskan Smoked Porter	6.1
Alaskan Brewing	Alaskan Winter Ale	6.2
Anchor	Anchor Steam	4.9	152
Anchor	Liberty Ale	6	188
Anchor	Anchor Porter	5.6	205


A bit of ruby:
rp = RowParser.new
data = rp.parse(File.open("beer.txt"))
data.each { |d|
	puts "#{d.beer}, #{d.calories}"
}


The output:
Amstel Light, 95
Alaskan Amber,
Alaskan Pale Ale,
Alaskan Stout,
Alaskan ESB,
Alaskan Smoked Porter,
Alaskan Winter Ale,
Anchor Steam, 152
Liberty Ale, 188
Anchor Porter, 205


Here's the source:

class String
	def conservative_split(delim)
		str = self.to_s
		bits = []
		while str.include? delim
			bits << str.slice!(0..str.index(delim)).strip
		end
		bits << str
		return bits
	end
end

class RowParser
	attr_reader :code

	def parse(data)
		# If it's an IO, read it in.  Else, assume it's a String.
		data = data.read if data.kind_of? IO
		data = data.split("\n") # Lines!

		# Create the Row class, based on the header columns
		headers = data.shift.split("\t")  # First line is the column names
		headers.collect! do |h|
			h = h.slice!(0).chr.downcase + h  # Downcase the first letter...
			h.gsub(/[^\w\d_]/, '')  # ...and purge problematic punctuation.
		end

		symbs = headers.collect { |h| ":#{h}" }.join(', ') # For attr_reader
		args = headers.join(', ') # For the constructor
		attrs = headers.collect { |h| "@#{h}" }.join(', ') # For the attrs
		@code = <<CODE
			class Row
				attr_reader #{symbs}
				def initialize(#{args})
					#{attrs} = #{args}
				end
				def to_s
					[#{attrs}].join(', ')
				end
			end
CODE
		eval @code

		# Now, create a Row from each remaining line
		rows = []

		data.each { |line|
			line = line.conservative_split("\t")
			rows << Row.new(*line)
		}
		
		return rows
	end
end

Hope it's useful...
Dan