From 0966889ab97dd00242a1db2da6d688a640375ce6 Mon Sep 17 00:00:00 2001
From: Raphael <raphael@joindiaspora.com>
Date: Tue, 4 Jan 2011 10:58:22 -0800
Subject: [PATCH] continue work on mongo export

---
 lib/mongo_to_mysql.rb           | 147 +++++++++++++++++++++++++++++++-
 lib/tasks/migrations.rake       |  22 +----
 spec/lib/mongo_to_mysql_spec.rb |  20 +++++
 3 files changed, 169 insertions(+), 20 deletions(-)
 create mode 100644 spec/lib/mongo_to_mysql_spec.rb

diff --git a/lib/mongo_to_mysql.rb b/lib/mongo_to_mysql.rb
index 2cc69d56c..974a08c4f 100644
--- a/lib/mongo_to_mysql.rb
+++ b/lib/mongo_to_mysql.rb
@@ -1,8 +1,34 @@
 #   Copyright (c) 2010, Diaspora Inc.  This file is
 #   licensed under the Affero General Public License version 3 or later.  See
 #   the COPYRIGHT file.
-
+require 'json'
+require 'csv'
 class MongoToMysql
+  def dirname
+    "tmp/export-for-mysql"
+  end
+  def dirpath
+    "#{Rails.root}/#{dirname}"
+  end
+  def make_dir
+    `mkdir -p #{dirname}/json`
+    `mkdir -p #{dirname}/csv`
+  end
+  def db_name
+    'diaspora-development'
+  end
+  def models
+    @models ||= [ {:name => :aspects},
+      {:name => :comments},
+      {:name => :contacts},
+      {:name => :invitations},
+      {:name => :notifications},
+      {:name => :people},
+      {:name => :posts},
+      {:name => :requests},
+      {:name => :users},
+    ]
+  end
   def id_sed
     @id_sed = sed_replace('{\ \"$oid\"\ :\ \(\"[^"]*\"\)\ }')
   end
@@ -12,4 +38,123 @@ class MongoToMysql
   def sed_replace(regex)
     "sed 's/#{regex}/\\1/g'"
   end
+  def json_for_model model_name
+    "mongoexport -d #{db_name} -c #{model_name} | #{id_sed} | #{date_sed}"
+  end
+  def write_json_export
+    models.each do |model|
+      filename ="#{dirpath}/json/#{model[:name]}.json"
+      model[:json_file] = filename
+      `#{json_for_model(model[:name])} > #{filename}`
+      debug "#{model[:name]} exported to #{dirname}/json/#{model[:name]}.json"
+    end
+    debug "Json export complete."
+  end
+  def debug string
+    if ['development', 'production'].include?(Rails.env)
+      puts string
+    end
+    Rails.logger.debug(string) if Rails.logger
+  end
+  def convert_json_files
+    models.each do |model|
+      self.send("#{model[:name]}_json_to_csv".to_sym, model)
+    end
+  end
+  def generic_json_to_csv model_hash
+    debug "Converting #{model_hash[:name]} json to csv"
+    json_file = File.open(model_hash[:json_file])
+
+    csv = CSV.open("#{dirpath}/csv/#{model_hash[:name]}.csv",'w')
+    csv << model_hash[:attrs]
+
+    json_file.each do |aspect_json|
+      hash = JSON.parse(aspect_json)
+      csv << yield(hash)
+    end
+    json_file.close
+    csv.close
+  end
+  def comments_json_to_csv model_hash
+    model_hash[:attrs] = ["mongo_id", "post_mongo_id", "person_mongo_id", "diaspora_handle", "text"]
+    generic_json_to_csv(model_hash) do |hash|
+      [hash["_id"], hash["post_id"], hash["person_id"], hash["diaspora_handle"], hash["text"]]
+    end
+  end
+  def contacts_json_to_csv model_hash
+    model_hash[:main_attrs] = ["mongo_id", "user_mongo_id", "person_mongo_id", "pending", "created_at", "updated_at"]
+    #Post Visibilities
+    model_hash[:join_table_name] = :aspect_memberships
+    model_hash[:join_table_attrs] = ["contact_mongo_id", "aspect_mongo_id"]
+
+    generic_json_to_two_csvs(model_hash) do |hash|
+      main_row = [hash["_id"], hash["user_id"], hash["person_id"], hash["pending"], hash["created_at"], hash["updated_at"]]
+      aspect_membership_rows = hash["aspect_ids"].map{|id| [hash["_id"],id]}
+      [main_row, aspect_membership_rows]
+    end
+    #Also writes the aspect memberships csv
+  end
+  def invitations_json_to_csv model_hash
+    model_hash[:attrs] = ["mongo_id", "recipient_mongo_id", "sender_mongo_id", "aspect_mongo_id", "message"]
+    generic_json_to_csv(model_hash) do |hash|
+      [hash["_id"], hash["to_id"], hash["from_id"], hash["into_id"], hash["message"]]
+    end
+  end
+  def notifications_json_to_csv model_hash
+    model_hash[:attrs] = ["mongo_id", "target_id", "target_type", "unread"]
+    generic_json_to_csv(model_hash) do |hash|
+      [hash["_id"], hash["target_id"], hash["kind"], hash["unread"]]
+    end
+  end
+  def people_json_to_csv model_hash
+    #Also writes the profiles csv
+  end
+  def posts_json_to_csv model_hash
+    #has to handle the polymorphic stuff
+  end
+  def requests_json_to_csv model_hash
+    model_hash[:attrs] = ["mongo_id", "recipient_mongo_id", "sender_mongo_id", "aspect_mongo_id"]
+    generic_json_to_csv(model_hash) do |hash|
+      [hash["_id"], hash["to_id"], hash["from_id"], hash["into_id"]]
+    end
+  end
+  def users_json_to_csv model_hash
+    model_hash[:attrs] = ["mongo_id", "username", "serialized_private_key", "encrypted_password", "invites", "invitation_token", "invitation_sent_at", "getting_started", "disable_mail", "language", "last_sign_in_ip", "last_sign_in_at", "reset_password_token", "password_salt"]
+    generic_json_to_csv(model_hash) do |hash|
+      [hash["_id"], hash["username"], hash["serialized_private_key"], hash["encrypted_password"], hash["invites"], hash["invitation_token"], hash["invitation_sent_at"], hash["getting_started"], hash["disable_mail"], hash["language"], hash["last_sign_in_ip"], hash["last_sign_in_at"], hash["reset_password_token"], hash["password_salt"]]
+    end
+  end
+  def aspects_json_to_csv model_hash
+    debug "Converting aspects json to aspects and post_visibilities csvs"
+    model_hash[:main_attrs] = ["mongo_id", "name", "created_at", "updated_at"]
+    #Post Visibilities
+    model_hash[:join_table_name] = :post_visibilities
+    model_hash[:join_table_attrs] = ["aspect_mongo_id", "post_mongo_id"]
+
+    generic_json_to_two_csvs(model_hash) do |hash|
+      main_row = [hash["_id"], hash["name"], hash["created_at"], hash["updated_at"]]
+      post_visibility_rows = hash["post_ids"].map{|id| [hash["_id"],id]}
+      [main_row, post_visibility_rows]
+    end
+  end
+  def generic_json_to_two_csvs model_hash
+    debug "Converting #{model_hash[:name]} json to two csvs"
+    json_file = File.open(model_hash[:json_file])
+
+    main_csv = CSV.open("#{dirpath}/csv/#{model_hash[:name]}.csv",'w')
+    main_csv << model_hash[:main_attrs]
+
+    join_csv = CSV.open("#{dirpath}/csv/#{model_hash[:join_table_name]}.csv",'w')
+    join_csv << model_hash[:join_table_attrs]
+
+    json_file.each do |aspect_json|
+      hash = JSON.parse(aspect_json)
+      result = yield(hash)
+      main_csv << result.first
+      result.last.each{|row| join_csv << row}
+    end
+    json_file.close
+    main_csv.close
+    join_csv.close
+  end
 end
diff --git a/lib/tasks/migrations.rake b/lib/tasks/migrations.rake
index df11d87f3..b795c3c9c 100644
--- a/lib/tasks/migrations.rake
+++ b/lib/tasks/migrations.rake
@@ -10,24 +10,8 @@ namespace :migrations do
   task :export_for_mysql do
     require 'lib/mongo_to_mysql'
     migrator = MongoToMysql.new
-    db_name = "diaspora-development"
-    models = [
-      :aspects,
-      :comments,
-      :contacts,
-      :invitations,
-      :notifications,
-      :people,
-      :posts,
-      :requests,
-      :users,
-    ]
-    `mkdir -p #{Rails.root}/tmp/export-for-mysql`
-    models.each do |model|
-      filename = "#{Rails.root}/tmp/export-for-mysql/#{model}.json"
-      `mongoexport -d #{db_name} -c #{model} | #{migrator.id_sed} | #{migrator.date_sed} > #{filename}`
-      puts "#{model} exported to #{filename}"
-      #`mongoexport -d #{db_name} -c #{model} -jsonArray | sed 's/\"[^"]*\"/"IAMID"/g' > #{filename}`
-    end
+    migrator.make_dir
+    migrator.write_json_export
+    migrator.convert_json_files
   end
 end
diff --git a/spec/lib/mongo_to_mysql_spec.rb b/spec/lib/mongo_to_mysql_spec.rb
new file mode 100644
index 000000000..8ecc73c84
--- /dev/null
+++ b/spec/lib/mongo_to_mysql_spec.rb
@@ -0,0 +1,20 @@
+require 'spec_helper'
+require 'lib/mongo_to_mysql'
+describe MongoToMysql do
+  before do
+    @migrator = MongoToMysql.new
+  end
+  describe '#sed_replace' do
+    before do
+      @test_string = '{ "_id" : { "$oid" : "4d0916c4cc8cb40e93000009" }, "name" : "Work", "created_at" : { "$date" : 1292441284000 }, "updated_at" : { "$date" : 1292546796000 }, "post_ids" : [ { "$oid" : "4d0aa87acc8cb4144b000009" }, { "$oid" : "4d0ab02ccc8cb41628000010" }, { "$oid" : "4d0ab2eccc8cb41628000011" } ], "user_id" : { "$oid" : "4d0916c2cc8cb40e93000006" } }'
+    end
+    it '#id_sed gets rid of the mongo id type specifier' do
+      post_sed = `echo '#{@test_string}' | #{@migrator.id_sed}`
+      post_sed.strip.match('"_id" : "4d0916c4cc8cb40e93000009", "name" : "Work",').should be_true
+    end
+    it '#date_sed gets rid of the mongo date type specifier' do
+      post_sed = `echo '#{@test_string}' | #{@migrator.date_sed}`
+      post_sed.strip.match('ork", "created_at" : 1292441284000, "updated_at" : 1292546796000, "post_ids" :').should be_true
+    end
+  end
+end