air18n 0.1.6 → 0.1.7

data/lib/air18n/backend.rb

@@ -165,7 +165,6 @@ module Air18n
         # the 'value' column reflects the latest English default text.
         store_default = true
         if result.present?
-          LoggingHelper.info "Default #{locale} text for key '#{key}' changed! Old default '#{result}' != new default '#{default}'. Route context: #{options[:routes_context]}"
           if @default_text_change_observer.present?
             @default_text_change_observer.default_text_changed(locale, key, result, default)
             store_default = @default_text_change_observer.allow_default_text_change?(locale, key, result, default)

data/lib/air18n/class_methods.rb

@@ -95,11 +95,13 @@ module Air18n
     end
 
     # Returns list of keys looked up today or in the last week.
+    # If we don't have data on phrase usage, returns all keys.
     def still_used_keys
       still_used_keys_hash.keys
     end
 
     # Returns list of phrase IDs that are still used.
+    # Returns empty list if we don't have data on phrase usage.
     def still_used_phrase_ids
       @still_used_phrase_ids ||= begin
         key_to_phrase_id = {}
@@ -115,6 +117,13 @@ module Air18n
       end
     end
 
+    # Returns whether or not a phrase key is still used.
+    # If we don't have data on phrase usage, then assumes all keys are still
+    # used.
+    def still_used?(key)
+      still_used_keys_hash.empty? || still_used_keys_hash.include?(key)
+    end
+
     # Phrases that are from user-generated content (like hosting names) have a
     # key prefixed with "ugc.".
     def phrase_key_is_ugc?(key)
@@ -125,7 +134,7 @@ module Air18n
     # A key needs to be translated if it is either UGC or was looked up by t()
     # today or in the last week.
     def needs_translation?(key)
-      phrase_key_is_ugc?(key) || still_used_keys_hash.include?(key)
+      phrase_key_is_ugc?(key) || still_used?(key)
     end
 
     # Create Air18n backend with given default_locale and translation prioritizer.

data/lib/air18n/phrase_translation.rb

@@ -122,95 +122,117 @@ module Air18n
       end
     end
 
-    # Helper method for translator_activity_data which counts words and payment
-    # of a set of translations and verifications, in (is verification bool) => (phrase id =>
-    # English phrase value map).
-    def self.construct_activity(phrase_ids_values_by_type)
-      {}.tap do |activity|
-        activity[:num_translations], activity[:word_count_translations] = translation_word_count(phrase_ids_values_by_type[false])
-        activity[:num_verifications], activity[:word_count_verifications] = translation_word_count(phrase_ids_values_by_type[true])
+    def self.translator_activity_data_new user_id=0, opts={}
+      user_criterion = user_id > 0 ? "AND pt.user_id=#{user_id}" : "AND NOT pt.user_id=0"
+      since_criterion = "AND pt.created_at >= '#{opts[:since].to_formatted_s(:db)}}'" if opts[:since]
+      sql = "SELECT year(pt.created_at) year, month(pt.created_at) month, pt.user_id user_id, pt.locale locale, day(pt.created_at) day, p.id phrase_id, pt.source_word_count source_word_count, pt.source_hash source_hash, pt.is_verification FROM phrase_translations pt, phrases p WHERE pt.phrase_id=p.id #{user_criterion} #{since_criterion} GROUP BY year(pt.created_at), month(pt.created_at), pt.user_id, pt.locale, pt.phrase_id, pt.source_hash, pt.is_verification"
+      res = self.connection.select_rows(sql)
+      phrases_per_user_locale_month_year =
+        Hash.new {|h, year| h[year] =
+          Hash.new {|h, month| h[month] =
+            Hash.new{|h, user_id| h[user_id] =
+              Hash.new{|h, locale| h[locale] =
+                Hash.new {|h, day| h[day] =
+                  Hash.new {|h, is_verification| h[is_verification] =
+                    Hash.new
+      } } } } } }
+
+      res.each do |row|
+        year, month, user_id, locale, day, phrase_id, source_word_count, source_hash, is_verification = row
+        is_verification = (is_verification == 1)
+        phrases_per_user_locale_month_year[year][month][user_id][locale][day][is_verification].merge!({source_hash => source_word_count})
       end
-    end
 
-    # Helper method for construct_activity which counts words in a set of
-    # translations or verifications, in phrase id => English phrase value map.
-    def self.translation_word_count(phrase_ids_to_values)
-      [phrase_ids_to_values.count, phrase_ids_to_values.values.join(' ').scan(/\w+/).size]
+      activities = []
+      phrases_per_user_locale_month_year.each do |year, months|
+        months.each do |month, user_ids|
+          user_ids.each do |user_id, locales|
+            locales.each do |locale, days|
+              if opts[:daily]
+                days.each do |day, phrase_ids_word_counts_by_type|
+                  activities << {:year => year, :month => month, :day => day, :locale => locale, :user_id => user_id, :activity => construct_activity_new(phrase_ids_word_counts_by_type)}
+                end
+              else
+                monthly_phrase_ids_word_counts_by_type = { false => {}, true => {} }
+                days.each do |_, phrase_ids_word_counts_by_type|
+                  [false, true].each do |is_verification|
+                    monthly_phrase_ids_word_counts_by_type[is_verification].merge!(phrase_ids_word_counts_by_type[is_verification])
+                  end
+                end
+                activity = construct_activity_new(monthly_phrase_ids_word_counts_by_type)
+                activities << {:year => year, :month => month, :locale => locale, :user_id => user_id, :activity => activity}
+              end
+            end
+          end
+        end
+      end
+      activities.sort do |a,b|
+        if a[:year] == b[:year]
+          if a[:month] == b[:month]
+            if a[:user_id] == b[:user_id]
+              if a[:locale] == b[:locale]
+                if !a[:day] || !b[:day] || a[:day] == b[:day]
+                  0
+                else
+                  a[:day] <=> b[:day]
+                end
+              else
+                a[:locale] <=> b[:locale]
+              end
+            else
+              a[:user_id] <=> b[:user_id]
+            end
+          else
+            a[:month] <=> b[:month]
+          end
+        else
+          a[:year] <=> b[:year]
+        end
+      end
     end
 
-    def self.activity_for_user_id uid, opts
-      translator_activity_data uid, opts
-    end
+    # Helper method for translator_activity_data which counts words and payment
+    # of a set of translations and verifications, in ((is verification bool => (phrase id =>
+    # source word count map)).
+    def self.construct_activity_new(phrase_ids_word_counts_by_type)
+      {}.tap do |activity|
+        activity[:num_translations] = phrase_ids_word_counts_by_type[false].size
+        activity[:word_count_translations] = phrase_ids_word_counts_by_type[false].values.sum
 
-    # Returns an array of integers representing user IDs who are not admins but
-    # who we also do not want to pay by the word for translations.
-    def self.fetch_unpaid_nonadmin_translator_list
-      rich_memory = RichMemory.find_by_memory_type(RichMemory::TYPE_TRANSLATOR_LIST)
-      if rich_memory
-        rich_memory.payload
-      else
-        []
+        activity[:num_verifications] = phrase_ids_word_counts_by_type[true].size
+        activity[:word_count_verifications] = phrase_ids_word_counts_by_type[true].values.sum
       end
     end
 
-    # Adds or removes a user from the unpaid-non-admin translator list.
-    def self.set_nonadmin_translator_payment_status id, should_be_paid
-      set = fetch_unpaid_nonadmin_translator_list.to_set
-      if should_be_paid
-        set -= [id]
-      else
-        set += [id]
-      end
-      write_unpaid_nonadmin_translator_list(set.to_a)
+    # Break text into words for purposes of word count.
+    # First uses to_s to convert text to a string; so for a nil input, returns
+    # empty array.
+    def self.segment(text)
+      text.to_s.scan(/\w+/)
     end
 
-    # translator_list: array of integers representing user IDs.
-    # Throws an exception if saving fails.
-    def self.write_unpaid_nonadmin_translator_list translator_list
-      rich_memory = RichMemory.find_by_memory_type(RichMemory::TYPE_TRANSLATOR_LIST) || RichMemory.new(:memory_type => RichMemory::TYPE_TRANSLATOR_LIST)
-      rich_memory.payload = translator_list
-      rich_memory.save!
+    def self.activity_for_user_id uid, opts
+      translator_activity_data uid, opts
     end
 
-    def self.fill_in_payment_data(activity_data, compute_payment_method)
-      activity_data.each do |stats|
-        stats[:currency] = 'EUR'
-        stats[:native_amount] = compute_payment_method.call(stats[:activity][:word_count_translations], stats[:activity][:word_count_verifications], stats[:user_id], stats[:locale].to_sym)
+    # Helper method for translator_activity_data which counts words and payment
+    # of a set of translations and verifications, in (is verification bool) => (phrase id =>
+    # English phrase value map).
+    #
+    # TODO(jkb) remove in September 2012.
+    def self.construct_activity(phrase_ids_values_by_type)
+      {}.tap do |activity|
+        activity[:num_translations], activity[:word_count_translations] = translation_word_count(phrase_ids_values_by_type[false])
+        activity[:num_verifications], activity[:word_count_verifications] = translation_word_count(phrase_ids_values_by_type[true])
       end
     end
 
-    def self.translator_payment_data(compute_payment_method)
-      today_last_month = 1.month.ago
-      ret = []
-      activity_data = PhraseTranslation.translator_activity_data(0, :since => 2.months.ago)
-      fill_in_payment_data(activity_data, compute_payment_method)
-      activity_data.each do |stats|
-        if stats[:year] == today_last_month.year && stats[:month] == today_last_month.month && stats[:user_id] > 0
-          user = User.find_by_id(stats[:user_id])
-          stats[:smart_name] = user.try(:smart_name)
-
-          stats[:is_admin] = user.try(:role_admin?)
-          stats[:is_paid] = !stats[:is_admin] && !fetch_unpaid_nonadmin_translator_list.include?(stats[:user_id])
-          stats[:usd_amount] = Currency.convert(stats[:native_amount], stats[:currency], "USD").round
-          if stats[:is_paid] && stats[:usd_amount] > 0
-            stats[:description] = "Translator payout "
-            stats[:description] << "for #{Date::MONTHNAMES[today_last_month.month]} #{today_last_month.year}. "
-            stats[:description] << "(#{stats[:activity][:word_count_translations]} translated words and #{stats[:activity][:word_count_verifications]} verified words in #{stats[:activity][:num_translations]} translated and #{stats[:activity][:num_verifications]} verified #{stats[:locale]} phrases.)"
-
-            stats[:line_item] = LineItem.new(
-              :user_id => stats[:user_id],
-              :ready_for_release_at => Date.today,
-              :sub_type => LineItem::SUB_TYPE_TRANSLATION_PAYOUT,
-              :native_currency => stats[:currency],
-              :amount => stats[:usd_amount],
-              :item => nil,
-              :description => (stats[:description] + " Thank you!!")
-            )
-          end
-          ret << stats
-        end
-      end
-      ret
+    # Helper method for construct_activity which counts words in a set of
+    # translations or verifications, in phrase id => English phrase value map.
+    #
+    # TODO(jkb) remove in September 2012.
+    def self.translation_word_count(phrase_ids_to_values)
+      [phrase_ids_to_values.count, segment(phrase_ids_to_values.values.join(' ')).size]
     end
 
     # Returns all translations for a locale. filter_opts are passed on to
@@ -236,7 +258,7 @@ module Air18n
       if filter_opts[:exclude_ugc] && I18n.phrase_key_is_ugc?(key)
         return false
       end
-      if filter_opts[:exclude_unused] && !I18n.still_used_keys_hash.include?(key)
+      if filter_opts[:exclude_unused] && !I18n.still_used?(key)
         return false
       end
       if filter_opts[:exclude_all]

data/lib/air18n/priority.rb

@@ -6,6 +6,9 @@ module Air18n
 
     # Returns a hash mapping translation keys to number of times they were
     # recently used.
+    #
+    # An empty hash means that we don't have data and should assume that all
+    # keys are still used.
     def key_usage
       {}
     end

data/lib/air18n/version.rb

@@ -1,3 +1,3 @@
 module Air18n
-  VERSION = "0.1.6"
+  VERSION = "0.1.7"
 end

data/lib/generators/air18n/migration/templates/active_record/migration.rb

@@ -28,6 +28,8 @@ class Air18nMigration < ActiveRecord::Migration
       t.boolean  "is_verification",                     :default => false
       t.boolean  "is_stale",                            :default => false
       t.boolean  "is_latest",                           :default => false
+      t.integer  "source_word_count",                   :default => 0
+      t.string   "source_hash"
     end
 
     add_index "phrase_translations", ["locale", "is_latest", "is_stale", "is_verification"], :name => "index_phrase_translations_on_locale_and_flags"

data/make_gem

@@ -1,5 +1,5 @@
 # Run this to push the new version of the gem to rubygems!
-v=0.1.6
+v=0.1.7
 gem build air18n.gemspec
 gem push air18n-${v}.gem
 

data/spec/lib/air18n/air18n_spec.rb

@@ -40,6 +40,11 @@ describe Air18n do
   end
 
   context 'Keeping track of priority' do
+    it 'should assume all phrases are used by default' do
+      I18n.still_used_keys_hash.should == {}
+      I18n.still_used?('magical new key').should == true
+    end
+
     it 'should call key_used' do
       Air18n::MockPriority.mock_priority({}) do |mock_priority|
         I18n.t('priority, key 1', :default => 'value 1')

data/spec/lib/air18n/phrase_translation_spec.rb

@@ -1,3 +1,5 @@
+# encoding: utf-8
+
 require 'spec_helper'
 
 describe Air18n::PhraseTranslation do
@@ -34,4 +36,13 @@ describe Air18n::PhraseTranslation do
     end
   end
 
+  context 'helpers' do
+    it 'should be able to segment' do
+      Air18n::PhraseTranslation.segment('Good check-in dates.').should == ['Good', 'check', 'in', 'dates']
+
+      # Ugh.
+      Air18n::PhraseTranslation.segment('こんにちは、ジェーソン。お元気ですか？').should == []
+    end
+  end
+
 end

data/spec/spec_helper.rb

@@ -60,3 +60,21 @@ def clean_database!
 end
 
 clean_database!
+
+# If you want to use created_at or updated_at in a test, pass your code as a
+# block to this method and the class that you model that you want to be able to
+# manually timestamp as an argument.
+#
+# Idea and implementation from:
+# http://codetunes.com/2009/05/09/turning-off-auto-timestamping-for-testing-in-rails
+def without_timestamping_of(*klasses)
+  if block_given?
+    klasses.delete_if { |klass| !klass.record_timestamps }
+    klasses.each { |klass| klass.record_timestamps = false }
+    begin
+      yield
+    ensure
+      klasses.each { |klass| klass.record_timestamps = true }
+    end
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: air18n
 version: !ruby/object:Gem::Version
-  version: 0.1.6
+  version: 0.1.7
   prerelease: 
 platform: ruby
 authors:
@@ -13,7 +13,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-07-26 00:00:00.000000000 Z
+date: 2012-08-30 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: i18n