bulk revert

app/logical/bulk_revert.rb

@@ -1,6 +1,17 @@
 class BulkRevert
+  BIG_QUERY_LIMIT = 5_000
   attr_reader :constraints
 
+  class ConstraintTooGeneralError < Exception ; end
+
+  def self.process(constraints)
+    obj = BulkRevert.new(constraints)
+
+    obj.find_post_versions.order("updated_at, id").each do |version|
+      version.undo!
+    end
+  end
+
   def initialize(constraints)
     @constraints = constraints
   end
@@ -18,14 +29,19 @@ class BulkRevert
 
     if constraints[:user_id]
       q = q.where("post_versions.updater_id = ?", constraints[:user_id])
+    end
 
-      if constraints[:added_tags] || constraints[:removed_tags]
-        hash = CityHash.hash64("#{constraints[:added_tags]} #{constraints{removed_tags}}").to_s(36)
-        sub_ids = Cache.get("br/fpv/#{hash}", 300) do
-          sub_ids = GoogleBigQuery::PostVersion.new.find(constraints[:user_id], constraints[:added_tags], constraints[:removed_tags])
-        end
-        q = q.where("post_versions.id in (?)", sub_ids)
+    if constraints[:added_tags] || constraints[:removed_tags]
+      hash = CityHash.hash64("#{constraints[:added_tags]} #{constraints{removed_tags}} #{constraints[:min_version_id]} #{constraints[:max_version_id]}").to_s(36)
+      sub_ids = Cache.get("br/fpv/#{hash}", 300) do
+        GoogleBigQuery::PostVersion.new.find(constraints[:user_id], constraints[:added_tags], constraints[:removed_tags], constraints[:min_version_id], constraints[:max_version_id], BIG_QUERY_LIMIT)
       end
+
+      if sub_ids.size >= BIG_QUERY_LIMIT
+        raise ConstraintTooGeneralError.new
+      end
+
+      q = q.where("post_versions.id in (?)", sub_ids)
     end
 
     if constraints[:min_version_id].present?

app/logical/google_big_query/post_version.rb

@@ -3,34 +3,50 @@ module GoogleBigQuery
     def find_removed(tag, limit = 1_000)
       tag = escape(tag)
       limit = limit.to_i
-      query("select id, post_id, updated_at, updater_id, updater_ip_addr, tags, added_tags, removed_tags, parent_id, rating, source from [#{data_set}.post_versions] where regexp_match(removed_tags, \"(?:^| )#{tag}(?:$| )\") order by updated_at desc limit #{limit}")
+      query("select id, post_id, updated_at, updater_id, updater_ip_addr, tags, added_tags, removed_tags, parent_id, rating, source from [#{data_set}.post_versions] where #{remove_tag_condition(tag)} order by updated_at desc limit #{limit}")
     end
 
     def find_added(tag, limit = 1_000)
       tag = escape(tag)
       limit = limit.to_i
-      query("select id, post_id, updated_at, updater_id, updater_ip_addr, tags, added_tags, removed_tags, parent_id, rating, source from [#{data_set}.post_versions] where regexp_match(added_tags, \"(?:^| )#{tag}(?:$| )\") order by updated_at desc limit #{limit}")
+      query("select id, post_id, updated_at, updater_id, updater_ip_addr, tags, added_tags, removed_tags, parent_id, rating, source from [#{data_set}.post_versions] where #{add_tag_condition(tag)} order by updated_at desc limit #{limit}")
     end
 
-    def find(user_id, added_tags, removed_tags, limit = 1_000)
+    def add_tag_condition(t)
+      es = escape(t)
+      "regexp_match(added_tags, \"(?:^| )#{es}(?:$| )\")"
+    end
+
+    def remove_tag_condition(t)
+      es = escape(t)
+      "regexp_match(removed_tags, \"(?:^| )#{es}(?:$| )\")"
+    end
+
+    def find(user_id, added_tags, removed_tags, min_version_id, max_version_id, limit = 1_000)
       constraints = []
 
       constraints << "updater_id = #{user_id.to_i}"
 
       if added_tags
         added_tags.scan(/\S+/).each do |tag|
-          escaped = escape(tag)
-          constraints << "regexp_match(added_tags, \"(?:^| )#{escaped}(?:$| )\")"
+          constraints << add_tag_condition(tag)
         end
       end
 
       if removed_tags
         removed_tags.scan(/\S+/).each do |tag|
-          escaped = escape(tag)
-          constraints << "not regexp_match(added_tags, \"(?:^| )#{escaped}(?:$| )\")"
+          constraints << remove_tag_condition(tag)
         end
       end
 
+      if min_version_id
+        constraints << "id >= #{min_version_id.to_i}"
+      end
+
+      if max_version_id
+        constraints << "id <= #{max_version_id.to_i}"
+      end
+
       limit = limit.to_i
       sql = "select id from [#{data_set}.post_versions] where " + constraints.join(" and ") + " order by updated_at desc limit #{limit}"
       result = query(sql)