From 28cbebe5ed23d004ebebc1d69e870e517402a5c7 Mon Sep 17 00:00:00 2001
From: Robin Ward <robin.ward@gmail.com>
Date: Thu, 2 Oct 2014 13:15:50 -0400
Subject: [PATCH] Better import for Drupal QA

---
 script/import_scripts/drupal.rb    |  6 +++++-
 script/import_scripts/drupal_qa.rb | 27 ++++++++++++++++++++++++---
 2 files changed, 29 insertions(+), 4 deletions(-)

diff --git a/script/import_scripts/drupal.rb b/script/import_scripts/drupal.rb
index 9db4503dfe5..7194a9a0fb5 100644
--- a/script/import_scripts/drupal.rb
+++ b/script/import_scripts/drupal.rb
@@ -17,6 +17,10 @@ class ImportScripts::Drupal < ImportScripts::Base
     )
   end
 
+  def categories_query
+    @client.query("SELECT tid, name, description FROM taxonomy_term_data WHERE vid = 1")
+  end
+
   def execute
     create_users(@client.query("SELECT uid id, name, mail email, created FROM users;")) do |row|
       {id: row['id'], username: row['name'], email: row['email'], created_at: Time.zone.at(row['created'])}
@@ -27,7 +31,7 @@ class ImportScripts::Drupal < ImportScripts::Base
     #   * Drupal allows duplicate category names, so you may need to exclude some categories or rename them here.
     #   * Table name may be term_data.
     #   * May need to select a vid other than 1.
-    create_categories(@client.query("SELECT tid, name, description FROM taxonomy_term_data WHERE vid = 1;")) do |c|
+    create_categories(categories_query) do |c|
       {id: c['tid'], name: c['name'], description: c['description']}
     end
 
diff --git a/script/import_scripts/drupal_qa.rb b/script/import_scripts/drupal_qa.rb
index d86e912cab6..6542812c0af 100644
--- a/script/import_scripts/drupal_qa.rb
+++ b/script/import_scripts/drupal_qa.rb
@@ -5,7 +5,28 @@ require "mysql2"
 
 class ImportScripts::DrupalQA < ImportScripts::Drupal
 
+  def categories_query
+    result = @client.query("SELECT n.nid, GROUP_CONCAT(ti.tid) AS tids
+                            FROM node AS n
+                            INNER JOIN taxonomy_index AS ti ON ti.nid = n.nid
+                            WHERE n.type = 'question'
+                              AND n.status = 1
+                            GROUP BY n.nid")
+
+    categories = {}
+    result.each do |r|
+      tids = r['tids']
+      if tids.present?
+        tids = tids.split(',')
+        categories[tids[0].to_i] = true
+      end
+    end
+
+    @client.query("SELECT tid, name, description FROM taxonomy_term_data WHERE tid IN (#{categories.keys.join(',')})")
+  end
+
   def create_forum_topics
+
     puts '', "creating forum topics"
 
     total_count = @client.query("
@@ -20,12 +41,12 @@ class ImportScripts::DrupalQA < ImportScripts::Drupal
       results = @client.query("
         SELECT n.nid,
                n.title,
-               MIN(t.field_question_tags_tid) AS tid,
+               GROUP_CONCAT(t.tid) AS tid,
                n.uid,
                n.created,
                f.body_value AS body
         FROM node AS n
-          LEFT OUTER JOIN field_data_field_question_tags AS t on t.entity_id = n.nid
+          LEFT OUTER JOIN taxonomy_index AS t on t.nid = n.nid
           INNER JOIN field_data_body AS f ON f.entity_id = n.nid
         WHERE n.type = 'question'
           AND n.status = 1
@@ -40,7 +61,7 @@ class ImportScripts::DrupalQA < ImportScripts::Drupal
         {
           id: "nid:#{row['nid']}",
           user_id: user_id_from_imported_user_id(row['uid']) || -1,
-          category: category_from_imported_category_id(row['tid']).try(:name),
+          category: category_from_imported_category_id((row['tid'] || '').split(',')[0]).try(:name),
           raw: row['body'],
           created_at: Time.zone.at(row['created']),
           pinned_at: nil,