markov chain generation is pretty decent

examples/word_count/markov

@@ -0,0 +1,34 @@
+#!/usr/bin/env ruby
+
+require '../../couchrest'
+
+cr = CouchRest.new("http://localhost:5984")
+db = cr.database('word-count-example')
+
+word = ARGV[0]
+words = [word]
+wprobs = {}
+
+while word
+  $stdout.print ' ' if words.length > 1
+  $stdout.print word
+  $stdout.flush
+  
+  wprobs[word] ||= db.view('markov/chain-reduce', :startkey => [word,nil], :endkey => [word,{}],:group_level => 2)
+
+  # puts
+  # puts "search #{word} #{wprobs[word]['rows'].length}"
+  # wprobs[word]['rows'].sort_by{|r|r['value']}.each{|r|puts [r['value'],r['key']].inspect}
+
+  rows = wprobs[word]['rows'].select{|r|(r['key'][1]!='')}.sort_by{|r|r['value']}
+  row = rows[(-1*[rows.length,5].min)..-1].sort_by{rand}[0]
+  word = row ? row['key'][1] : nil
+  words << word
+end
+
+$stdout.print '.'
+$stdout.flush
+puts
+
+# `say #{words.join(' ')}`
+

examples/word_count/views/books/chunked-map.js

@@ -0,0 +1,3 @@
+function(doc) {
+  doc.title && doc.chunk && emit([doc.title, doc.chunk],null);
+}

examples/word_count/views/books/united-map.js

@@ -0,0 +1 @@
+function(doc){if(doc.text && doc.text.match(/united/))  emit([doc.title, doc.chunk],null)}

examples/word_count/views/markov/chain-map.js

@@ -0,0 +1,6 @@
+function(doc){
+  var words = doc.text.split(/\W/).filter(function(w) {return w.length > 0}).map(function(w){return w.toLowerCase()});
+  for (var i = 0, l = words.length; i < l; i++) {
+    emit(words.slice(i,4),doc.title);
+  }
+}

examples/word_count/views/markov/chain-reduce.js

@@ -0,0 +1,7 @@
+function(key,vs,c){
+  if (c) {
+    return sum(vs);
+  } else {
+    return vs.length;
+  }
+}

examples/word_count/views/word_count/count-map.js

@@ -0,0 +1,6 @@
+function(doc){
+  var words = doc.text.split(/\W/).map(function(w){return w.toLowerCase()});
+  words.forEach(function(word){
+    if (word.length > 0) emit([word,doc.title],1);
+  });
+}

examples/word_count/views/word_count/count-reduce.js

@@ -0,0 +1,3 @@
+function(key,combine){
+  return sum(combine);
+}

examples/word_count/word_count.rb

@@ -2,16 +2,18 @@ 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')
 
 %w{america.txt da-vinci.txt outline-of-science.txt ulysses.txt}.each do |book|
+# %w{}.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
+      puts chunk
       lines << line
       if lines.length > 10
         db.save({