From d8d5645ebd0b457633d27b469a96685fc016dccd Mon Sep 17 00:00:00 2001
From: Nolan Darilek <nolan@thewordnerd.info>
Date: Sun, 14 Dec 2008 23:17:35 -0600
Subject: [PATCH] Make bulk saving more flexible.  * Database#save(doc, true)
 caches the doc in a database-specific bulk cache. * Database#save(doc,
 false), default, saves normally, bulk saving and emptying the cache if one
 exists. * The cache is automatically saved on Database#save if it excedes a
 configurable limit, 50 by default. * Database#bulk_save without arguments
 saves and clears the bulk save cache.

---
 lib/couchrest/core/database.rb       | 23 ++++++++++--
 spec/couchrest/core/database_spec.rb | 53 +++++++++++++++++++++++++++-
 2 files changed, 73 insertions(+), 3 deletions(-)

diff --git a/lib/couchrest/core/database.rb b/lib/couchrest/core/database.rb
index 184d362..6aecf70 100644
--- a/lib/couchrest/core/database.rb
+++ b/lib/couchrest/core/database.rb
@@ -4,6 +4,7 @@ require "base64"
 module CouchRest
   class Database
     attr_reader :server, :host, :name, :root
+    attr_accessor :bulk_save_cache_limit
      
     # Create a CouchRest::Database adapter for the supplied CouchRest::Server
     # and database name.
@@ -18,6 +19,8 @@ module CouchRest
       @host = server.uri
       @root = "#{host}/#{name}"
       @streamer = Streamer.new(self)
+      @bulk_save_cache = []
+      @bulk_save_cache_limit = 50
     end
     
     # returns the database's uri
@@ -106,10 +109,20 @@ module CouchRest
     # documents on the client side because POST has the curious property of
     # being automatically retried by proxies in the event of network
     # segmentation and lost responses.
-    def save doc
+    #
+    # If <tt>bulk</tt> is true (false by default) the document is cached for bulk-saving later.
+    # Bulk saving happens automatically when #bulk_save_cache limit is exceded, or on the next non bulk save.
+    def save (doc, bulk = false)
       if doc['_attachments']
         doc['_attachments'] = encode_attachments(doc['_attachments'])
       end
+      if bulk
+        @bulk_save_cache << doc
+        return bulk_save if @bulk_save_cache.length >= @bulk_save_cache_limit
+        return doc
+      elsif !bulk && @bulk_save_cache.length > 0
+        bulk_save
+      end
       result = if doc['_id']
         slug = CGI.escape(doc['_id'])
         CouchRest.put "#{@root}/#{slug}", doc
@@ -131,7 +144,13 @@ module CouchRest
     
     # POST an array of documents to CouchDB. If any of the documents are
     # missing ids, supply one from the uuid cache.
-    def bulk_save docs
+    #
+    # If called with no arguments, bulk saves the cache of documents to be bulk saved.
+    def bulk_save (docs = nil)
+      if docs.nil?
+        docs = @bulk_save_cache
+        @bulk_save_cache = []
+      end
       ids, noids = docs.partition{|d|d['_id']}
       uuid_count = [noids.length, @server.uuid_batch_count].max
       noids.each do |doc|
diff --git a/spec/couchrest/core/database_spec.rb b/spec/couchrest/core/database_spec.rb
index 1be153c..f30ef06 100644
--- a/spec/couchrest/core/database_spec.rb
+++ b/spec/couchrest/core/database_spec.rb
@@ -190,6 +190,16 @@ describe CouchRest::Database do
         @db.get('twoB')
       end.should raise_error(RestClient::ResourceNotFound)
     end
+
+    it "should empty the bulk save cache if no documents are given" do
+      @db.save({"_id" => "bulk_cache_1", "val" => "test"}, true)
+      lambda do
+        @db.get('bulk_cache_1')
+      end.should raise_error(RestClient::ResourceNotFound)
+      @db.bulk_save
+      @db.get("bulk_cache_1")["val"].should == "test"
+    end
+
     it "should raise an error that is useful for recovery" do
       @r = @db.save({"_id" => "taken", "field" => "stuff"})
       begin
@@ -400,7 +410,48 @@ describe CouchRest::Database do
       now['them-keys'].should == 'huge'
     end
   end
-  
+
+  describe "cached bulk save" do
+    it "stores documents in a database-specific cache" do
+      td = {"_id" => "btd1", "val" => "test"}
+      @db.save(td, true)
+      @db.instance_variable_get("@bulk_save_cache").should == [td]
+      
+    end
+
+    it "doesn't save to the database until the configured cache size is exceded" do
+      @db.bulk_save_cache_limit = 3
+      td1 = {"_id" => "td1", "val" => true}
+      td2 = {"_id" => "td2", "val" => 4}
+      @db.save(td1, true)
+      @db.save(td2, true)
+      lambda do
+        @db.get(td1["_id"])
+      end.should raise_error(RestClient::ResourceNotFound)
+      lambda do
+        @db.get(td2["_id"])
+      end.should raise_error(RestClient::ResourceNotFound)
+      td3 = {"_id" => "td3", "val" => "foo"}
+      @db.save(td3, true)
+      @db.get(td1["_id"])["val"].should == td1["val"]
+      @db.get(td2["_id"])["val"].should == td2["val"]
+      @db.get(td3["_id"])["val"].should == td3["val"]
+    end
+
+    it "clears the bulk save cache the first time a non bulk save is requested" do
+      td1 = {"_id" => "blah", "val" => true}
+      td2 = {"_id" => "steve", "val" => 3}
+      @db.bulk_save_cache_limit = 50
+      @db.save(td1, true)
+      lambda do
+        @db.get(td1["_id"])
+      end.should raise_error(RestClient::ResourceNotFound)
+      @db.save(td2)
+      @db.get(td1["_id"])["val"].should == td1["val"]
+      @db.get(td2["_id"])["val"].should == td2["val"]
+    end
+  end
+
   describe "DELETE existing document" do
     before(:each) do
       @r = @db.save({'lemons' => 'from texas', 'and' => 'spain'})