FIX: First Quote badge bust

Feature: track quoted posts

app/controllers/posts_controller.rb

@@ -123,6 +123,7 @@ def update
     revisor = PostRevisor.new(post)
     if revisor.revise!(current_user, params[:post][:raw], edit_reason: params[:post][:edit_reason])
       TopicLink.extract_from(post)
+      QuotedPost.extract_from(post)
     end
 
     if post.errors.present?

app/models/badge.rb

@@ -45,17 +45,16 @@ module Queries
 SQL
 
     FirstQuote = <<SQL
-    SELECT l.user_id, l.post_id, l.created_at granted_at
+    SELECT ids.user_id, q.post_id, q.created_at granted_at
     FROM
     (
-      SELECT MIN(l1.id) id
-      FROM topic_links l1
-      JOIN badge_posts p1 ON p1.id = l1.post_id
-      JOIN badge_posts p2 ON p2.id = l1.link_post_id
-      WHERE NOT reflection AND quote
-      GROUP BY l1.user_id
+      SELECT p1.user_id, MIN(q1.id) id
+      FROM quoted_posts q1
+      JOIN badge_posts p1 ON p1.id = q1.post_id
+      JOIN badge_posts p2 ON p2.id = q1.quoted_post_id
+      GROUP BY p1.user_id
     ) ids
-    JOIN topic_links l ON l.id = ids.id
+    JOIN quoted_posts q ON q.id = ids.id
 SQL
 
     FirstLink = <<SQL

app/models/post.rb

@@ -93,6 +93,7 @@ def recover!
     super
     update_flagged_posts_count
     TopicLink.extract_from(self)
+    QuotedPost.extract_from(self)
     if topic && topic.category_id && topic.category
       topic.category.update_latest
     end
@@ -336,7 +337,9 @@ def rebake!(opts={})
     update_columns(cooked: new_cooked, baked_at: Time.new, baked_version: BAKED_VERSION)
 
     # Extracts urls from the body
-    TopicLink.extract_from self
+    TopicLink.extract_from(self)
+    QuotedPost.extract_from(self)
+
     # make sure we trigger the post process
     trigger_post_process(true)
 

app/models/quoted_post.rb

@@ -0,0 +1,46 @@
+class QuotedPost < ActiveRecord::Base
+  belongs_to :post
+  belongs_to :quoted_post, class_name: 'Post'
+
+  # NOTE we already have a path that does this for topic links,
+  #  however topic links exclude quotes and links within a topic
+  #  we are double parsing this fragment, this may be worth optimising later
+  def self.extract_from(post)
+
+    doc = Nokogiri::HTML.fragment(post.cooked)
+
+    uniq = {}
+    ids = []
+
+    doc.css("aside.quote[data-topic]").each do |a|
+      topic_id = a['data-topic'].to_i
+      post_number = a['data-post'].to_i
+
+      next if topic_id == 0 || post_number == 0
+      next if uniq[[topic_id,post_number]]
+      uniq[[topic_id,post_number]] = true
+
+
+      # It would be so much nicer if we used post_id in quotes
+      results = exec_sql "INSERT INTO quoted_posts(post_id, quoted_post_id, created_at, updated_at)
+               SELECT :post_id, id, current_timestamp, current_timestamp
+               FROM posts
+               WHERE post_number = :post_number AND
+                     topic_id = :topic_id
+               RETURNING quoted_post_id
+      ", post_id: post.id, post_number: post_number, topic_id: topic_id
+
+      results = results.to_a
+
+      if results.length > 0
+        ids << results[0]["quoted_post_id"].to_i
+      end
+    end
+
+    if ids.length > 0
+      exec_sql "DELETE FROM quoted_posts WHERE post_id = :post_id AND quoted_post_id NOT IN (:ids)",
+          post_id: post.id, ids: ids
+    end
+
+  end
+end

app/models/topic.rb

@@ -487,6 +487,7 @@ def add_moderator_post(user, text, opts={})
 
       # Grab any links that are present
       TopicLink.extract_from(new_post)
+      QuotedPost.extract_from(new_post)
     end
 
     new_post

db/migrate/20140715055242_add_quoted_posts.rb

@@ -0,0 +1,54 @@
+class AddQuotedPosts < ActiveRecord::Migration
+  def change
+    create_table :quoted_posts do |t|
+      t.integer :post_id, null: false
+      t.integer :quoted_post_id, null: false
+      t.timestamps
+    end
+
+    add_index :quoted_posts, [:post_id, :quoted_post_id], unique: true
+    add_index :quoted_posts, [:quoted_post_id, :post_id], unique: true
+
+
+    # NOTE this can be done in pg but too much of a headache
+    id = 0
+    while id = backfill_batch(id, 1000); end
+  end
+
+  def backfill_batch(start_id, batch_size)
+
+    results = execute <<SQL
+    SELECT id, cooked
+    FROM posts
+    WHERE raw like '%quote=%' AND id > #{start_id}
+    ORDER BY id
+    LIMIT #{batch_size}
+SQL
+
+    max_id = nil
+
+    results.each do |row|
+      post_id, max_id = row["id"].to_i
+      doc = Nokogiri::HTML.fragment(row["cooked"])
+
+      uniq = {}
+
+      doc.css("aside.quote[data-topic]").each do |a|
+        topic_id = a['data-topic'].to_i
+        post_number = a['data-post'].to_i
+
+        next if uniq[[topic_id,post_number]]
+        uniq[[topic_id,post_number]] = true
+
+
+        execute "INSERT INTO quoted_posts(post_id, quoted_post_id, created_at, updated_at)
+                 SELECT #{post_id}, id, created_at, updated_at
+                 FROM posts
+                 WHERE post_number = #{post_number} AND
+                       topic_id = #{topic_id}"
+      end
+    end
+
+    max_id
+  end
+end

lib/post_creator.rb

@@ -266,6 +266,7 @@ def publish
 
   def extract_links
     TopicLink.extract_from(@post)
+    QuotedPost.extract_from(@post)
   end
 
   def track_topic

spec/models/quoted_post_spec.rb

@@ -0,0 +1,27 @@
+require 'spec_helper'
+
+describe QuotedPost do
+  it 'correctly extracts quotes in integration test' do
+    post1 = create_post
+    post2 = create_post(topic_id: post1.topic_id,
+                        raw: "[quote=\"#{post1.user.username}, post: 1, topic:#{post1.topic_id}\"]\ntest\n[/quote]\nthis is a test post")
+
+    QuotedPost.find_by(post_id: post2.id, quoted_post_id: post1.id).should_not be_nil
+  end
+
+  it 'correctly handles deltas' do
+    post1 = Fabricate(:post)
+    post2 = Fabricate(:post)
+
+    post2.cooked = <<HTML
+<aside class="quote" data-post="#{post1.post_number}" data-topic="#{post1.topic_id}"><div class="title"><div class="quote-controls"></div><img width="20" height="20" src="/user_avatar/meta.discourse.org/techapj/20/3281.png" class="avatar">techAPJ said:</div><blockquote><p>When the user will v</p></blockquote></aside>
+HTML
+
+    QuotedPost.create!(post_id: post2.id, quoted_post_id: 999)
+
+    QuotedPost.extract_from(post2)
+    QuotedPost.where(post_id: post2.id).count.should == 1
+    QuotedPost.find_by(post_id: post2.id, quoted_post_id: post1.id).should_not be_nil
+
+  end
+end