couchrest_model/examples/word_count/word_count.rb

56 lines
1.4 KiB
Ruby
Raw Normal View History

2008-05-25 20:49:37 +02:00
require File.dirname(__FILE__) + '/../../couchrest'
couch = CouchRest.new("http://localhost:5984")
db = couch.database('word-count-example')
db.delete! rescue nil
db = couch.create_db('word-count-example')
2008-05-25 20:49:37 +02:00
2008-06-01 19:08:38 +02:00
%w{america.txt da-vinci.txt outline-of-science.txt ulysses.txt}.each do |book|
# %w{}.each do |book|
2008-05-25 20:49:37 +02:00
title = book.split('.')[0]
puts title
File.open(File.join(File.dirname(__FILE__),book),'r') do |file|
lines = []
chunk = 0
while line = file.gets
puts chunk
2008-05-25 20:49:37 +02:00
lines << line
2008-06-01 19:08:38 +02:00
if lines.length > 10
2008-05-25 20:49:37 +02:00
db.save({
:title => title,
:chunk => chunk,
:text => lines.join('')
})
chunk += 1
lines = []
end
end
end
end
2008-06-01 19:08:38 +02:00
# word_count = {
# :map => 'function(doc){
# var words = doc.text.split(/\W/);
# words.forEach(function(word){
# if (word.length > 0) emit([word,doc.title],1);
# });
# }',
# :reduce => 'function(key,combine){
# return sum(combine);
# }'
# }
#
# db.delete db.get("_design/word_count") rescue nil
#
# db.save({
# "_id" => "_design/word_count",
# :views => {
# :count => word_count,
# :words => {:map => word_count[:map]}
# }
# })
2008-05-25 20:49:37 +02:00
2008-05-26 07:35:04 +02:00
puts "The books have been stored in your CouchDB. To initiate the MapReduce process, visit http://localhost:5984/_utils/ in your browser and click 'word-count-example', then select view 'words' or 'count'. The process could take about 15 minutes on an average MacBook."
2008-05-25 20:49:37 +02:00