word count smaller chunks

This commit is contained in:
Chris Anderson 2008-06-01 10:08:38 -07:00
parent f9c95bfce6
commit 32b8090ddb

View file

@ -2,10 +2,10 @@ 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')
# db.delete! rescue nil
# db = couch.create_db('word-count-example')
['da-vinci.txt', 'outline-of-science.txt', 'ulysses.txt'].each do |book|
%w{america.txt da-vinci.txt outline-of-science.txt ulysses.txt}.each do |book|
title = book.split('.')[0]
puts title
File.open(File.join(File.dirname(__FILE__),book),'r') do |file|
@ -13,7 +13,7 @@ db = couch.create_db('word-count-example')
chunk = 0
while line = file.gets
lines << line
if lines.length > 100
if lines.length > 10
db.save({
:title => title,
:chunk => chunk,
@ -26,27 +26,27 @@ db = couch.create_db('word-count-example')
end
end
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]}
}
})
# 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]}
# }
# })
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."