From fbccbd73e0bd4b00ce097a85d92696a49d701aca Mon Sep 17 00:00:00 2001
From: Chris Anderson <jchris@grabb.it>
Date: Sun, 3 Aug 2008 14:17:58 -0700
Subject: [PATCH] created some utility scripts

---
 utils/remap.rb  | 27 +++++++++++++++++++++++++++
 utils/subset.rb | 30 ++++++++++++++++++++++++++++++
 2 files changed, 57 insertions(+)
 create mode 100644 utils/remap.rb
 create mode 100644 utils/subset.rb

diff --git a/utils/remap.rb b/utils/remap.rb
new file mode 100644
index 0000000..f42c5d5
--- /dev/null
+++ b/utils/remap.rb
@@ -0,0 +1,27 @@
+require 'rubygems'
+require 'couchrest'
+
+# set the source db and map view
+source = CouchRest.new("http://localhost:5984").database('source-db')
+source_view = 'mydesign/view-map'
+
+# set the target db
+target = CouchRest.new("http://localhost:5984").database('target-db')
+
+
+pager = CouchRest::Pager.new(source)
+
+# pager will yield once per uniq key in the source view
+
+pager.key_reduce(source_view, 10000) do |key, values|
+  # create a doc from the key and the values
+  example_doc = {
+    :key => key,
+    :values => values.uniq
+  }
+
+  target.save(example_doc)
+  
+  # keep us up to date with progress
+  puts k if (rand > 0.9)
+end
diff --git a/utils/subset.rb b/utils/subset.rb
new file mode 100644
index 0000000..dd5d2e7
--- /dev/null
+++ b/utils/subset.rb
@@ -0,0 +1,30 @@
+require 'rubygems'
+require 'couchrest'
+
+# subset.rb replicates a percentage of a database to a fresh database.
+# use it to create a smaller dataset on which to prototype views.
+
+# specify the source database
+source = CouchRest.new("http://localhost:5984").database('source-db')
+
+# specify the target database
+target = CouchRest.new("http://localhost:5984").database('target-db')
+
+# pager efficiently yields all view rows
+pager = CouchRest::Pager.new(source)
+
+pager.all_docs(1000) do |rows|  
+  docs = rows.collect do |r|
+    # the percentage of docs to clone
+    next if rand > 0.1 
+    doc = source.get(r['id'])
+    doc.delete('_rev')
+    doc      
+  end.compact
+  puts docs.length
+  next if docs.empty?
+
+  puts docs.first['_id']
+  target.bulk_save(docs)
+end
+