word count examples

examples/word_count/word_count.rb

@@ -0,0 +1,53 @@
+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')
+
+['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|
+    lines = []
+    chunk = 0
+    while line = file.gets
+      lines << line
+      if lines.length > 100
+        db.save({
+          :title => title,
+          :chunk => chunk, 
+          :text => lines.join('')
+        })
+        chunk += 1
+        lines = []
+      end
+    end
+  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]}
+  }
+})
+
+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 around an hour on an average MacBook."
+
+

examples/word_count/word_count_query.rb

@@ -0,0 +1,34 @@
+require File.dirname(__FILE__) + '/../../couchrest'
+
+couch = CouchRest.new("http://localhost:5984")
+db = couch.database('word-count-example')
+
+puts "Now that we've parsed all those books into CouchDB, the queries we can run are incredibly flexible."
+
+puts "\nThe simplest query we can run is the total word count for all words in all documents:"
+puts db.view('word_count/count').inspect
+
+puts "\nWe can also narrow the query down to just one word, across all documents. Here is the count for 'flight':"
+word = 'flight'
+params = {
+  :startkey => [word], 
+  :endkey => [word,'Z']
+  }
+
+puts db.view('word_count/count',params).inspect
+
+puts "\nWe scope the query using startkey and endkey params to take advantage of CouchDB's collation ordering. Here are the params for the last query:"
+puts params.inspect
+
+puts "\nWe can also count words on a per-title basis."
+title = 'da-vinci'
+params = {
+  :key => [word, title]
+  }
+puts db.view('word_count/count',params).inspect
+puts "\nHere are the params for 'flight' in the da-vinci book:"
+puts params.inspect
+puts
+puts 'The url looks like this:'
+puts 'http://localhost:5984/word-count-example/_view/word_count/count?key=["flight","da-vinci"]'
+puts "\nTry dropping that in your browser..."