On Tuesday evening, I gave a 20 slide ‘lightning talk’ at the London Ruby Users Group (LRUG) entitled “Ruby Poetry”. Inspired by Murray Steele’s ”My First Ruby” talk at a previous LRUG, I decided to tell a story about a small progam I wrote about five years ago, when I was fairly new to Ruby. At the time, I had just made the transition to Ruby after years as Java developer, and I was amazed at how easy it was to solve problems without having to write very much code.

The story starts when I was invited to a party. The party was a themed Burns Night dinner. These celebrations are common in Scotland to celebrate the life and poetry of Robert Burns. I had been invited to the same party the previous year, and remembered back to enjoying the haggis and whisky. One thing I recalled that had not felt particularly comfortable with, is that I had been asked to bring a poem to read. On the evening, instead of bringing poems from books, most people had written their own, and I felt bad for lacking creativity. I decided that this time around, I would try to be creative in my own way, and wrote a Ruby program to generate some topical poetry from the day’s news headlines. I won’t go into too much detail here, as my five minute talk was kindly videoed by Skills Matter:

generate_topical_rhyming_couplets.rb

The code below is what I used to generate the couplets on the day. While I had been tuning my algorithm, I wrote the headlines to a file so I didn’t have to reload them each time. The full source is on github.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50

#!/usr/bin/env ruby
require 'rss'
require 'open-uri'

# Open the list of feeds
fh = open("feeds.txt")
feeds = fh.read.split "\n"

# For each feed, get the headlines
headlines = []
feeds.each do |feed|
  begin
    file = open(feed)
    rss = RSS::Parser.parse(file.read)
    feed_headlines = rss.items.map {|i| i.title}
    headlines += feed_headlines
  rescue
    puts "Error with #{feed}"
  end
end

# Remove duplicates and strip some prefixes
headlines = headlines.uniq!
headlines = headlines.map {|h| h.gsub /^(AUDIO|VIDEO): /,""}

# Test for couplets
def couplet?(a,b)
  as = a.gsub /\W/,"" # Ignore punctuation
  bs = b.gsub /\W/,""
  as[-3,3] == bs[-3,3] &&
    !(as[-5,5] == bs[-5,5]) &&
    (as.length - bs.length).abs < 10 &&
    (as.length > 30 && as.length < 70) &&
    (bs.length > 30 && bs.length < 70)
end

# Sort the headlines by the last three letters
headlines.sort! {|a,b|  a[-3,3] <=> b[-3,3] }

# Iterate over the headlines, testing in pairs for couplets
i = 0
while (i < headlines.length-2) do
  a = headlines[i]
  b = headlines[i+1]
  if couplet?(a,b)
    i += 1
    puts "#{a}\n#{b}\n\n"
  end
  i += 1
end

As the couplet finding algorithm was fairly crude, only looking for matches of the last three letters, and not the phonetic representation, I allowed myself to pick the best couplets as generated on the day. To demonstrate my code, I reran the program on the day of my talk, and read the best couplets it generated at the end of my presentation:

The final poem, as generated on the day of the talk

I particularly like the last one, which like the others, was genuinely generated on the day.

So the two talks I have given at LRUG have been entitled Ruby Poetry and Ruby Golf. What Ruby <insert word here> should I do next?

Update 1: Matt Westcott uses the generated couplets as lyrics to a song. Amazing: http://fawm.org/songs/29365

Update 2: Sebastian Spier forked the code, wrote some documentation, refactored it, and wrote a script to output HTML with headlines linking to the original news sources.

Tribesports is hiring developers in London. Find out more here.