From 59223b283a0de158bcfd2ef3262346efba16251e Mon Sep 17 00:00:00 2001
From: Gerhard Schlager <gerhard.schlager@discourse.org>
Date: Tue, 22 Jun 2021 21:00:51 +0200
Subject: [PATCH] WIP bulk importer

---
 Gemfile                            |  3 +
 Gemfile.lock                       |  6 +-
 script/bulk_import/generic_bulk.rb | 96 ++++++++++++++++++++++++++++++
 3 files changed, 104 insertions(+), 1 deletion(-)
 create mode 100644 script/bulk_import/generic_bulk.rb

diff --git a/Gemfile b/Gemfile
index e09839e88a..c493a03daf 100644
--- a/Gemfile
+++ b/Gemfile
@@ -258,3 +258,6 @@ gem 'colored2', require: false
 gem 'maxminddb'
 
 gem 'rails_failover', require: false
+
+gem 'redcarpet'
+gem "sqlite3"
diff --git a/Gemfile.lock b/Gemfile.lock
index 53803ed4d0..f9c736a972 100644
--- a/Gemfile.lock
+++ b/Gemfile.lock
@@ -345,6 +345,7 @@ GEM
       msgpack (>= 0.4.3)
       optimist (>= 3.0.0)
     rchardet (1.8.0)
+    redcarpet (3.5.1)
     redis (4.5.1)
     redis-namespace (1.8.1)
       redis (>= 3.0.4)
@@ -446,6 +447,7 @@ GEM
       actionpack (>= 4.0)
       activesupport (>= 4.0)
       sprockets (>= 3.0.0)
+    sqlite3 (1.4.2)
     sshkey (2.0.0)
     stackprof (0.2.17)
     test-prof (1.0.7)
@@ -576,6 +578,7 @@ DEPENDENCIES
   rb-fsevent
   rbtrace
   rchardet
+  redcarpet
   redis
   redis-namespace
   rinku
@@ -600,6 +603,7 @@ DEPENDENCIES
   simplecov
   sprockets (= 3.7.2)
   sprockets-rails
+  sqlite3
   sshkey
   stackprof
   test-prof
@@ -613,4 +617,4 @@ DEPENDENCIES
   yaml-lint
 
 BUNDLED WITH
-   2.2.26
+   2.2.29
diff --git a/script/bulk_import/generic_bulk.rb b/script/bulk_import/generic_bulk.rb
new file mode 100644
index 0000000000..19d34e2a91
--- /dev/null
+++ b/script/bulk_import/generic_bulk.rb
@@ -0,0 +1,96 @@
+# frozen_string_literal: true
+
+require_relative "base"
+require "sqlite3"
+
+class BulkImport::Generic < BulkImport::Base
+  AVATAR_DIRECTORY = ENV["AVATAR_DIRECTORY"]
+  UPLOAD_DIRECTORY = ENV["UPLOAD_DIRECTORY"]
+
+  def initialize(db_path)
+    super()
+    @db = create_connection(db_path)
+  end
+
+  def start
+    run # will call execute, and then "complete" the migration
+
+    # Now that the migration is complete, do some more work:
+
+    Discourse::Application.load_tasks
+    Rake::Task["import:ensure_consistency"].invoke
+  end
+
+  def execute
+    import_groups
+    import_users
+  end
+
+  def import_groups
+    puts "Importing groups..."
+
+    groups = @db.execute(<<~SQL, last_row_id: @last_imported_group_id)
+      SELECT *
+      FROM groups
+      WHERE ROWID > :last_row_id #{}
+      ORDER BY ROWID
+    SQL
+
+    create_groups(groups) do |row|
+      {
+        imported_id: row["id"],
+        name: normalize_text("name")
+      }
+    end
+  end
+
+  def import_users
+    puts "Importing users..."
+
+    users = @db.execute(<<~SQL, last_row_id: @last_imported_user_id)
+      SELECT ROWID, *
+      FROM users
+      WHERE ROWID > :last_row_id
+      ORDER BY ROWID
+    SQL
+
+    create_users(users) do |row|
+      {
+        imported_id: row["id"],
+        username: row["username"],
+        created_at: to_datetime(row["created_at"]),
+        name: row["name"],
+        email: row["email"],
+        last_seen_at: to_datetime(row[:"last_seen_at"]),
+        bio_raw: row["bio"],
+        location: row["location"],
+        admin: to_boolean(row["admin"]),
+        moderator: to_boolean(row["moderator"])
+      }
+    end
+  end
+
+  def create_connection(path)
+    sqlite = SQLite3::Database.new(path, results_as_hash: true)
+    sqlite.busy_timeout = 60000 # 60 seconds
+    sqlite.auto_vacuum = "full"
+    sqlite.foreign_keys = true
+    sqlite.journal_mode = "wal"
+    sqlite.synchronous = "normal"
+    sqlite
+  end
+
+  def to_date(text)
+    text.present? ? Date.parse(text) : nil
+  end
+
+  def to_datetime(text)
+    text.present? ? DateTime.parse(text) : nil
+  end
+
+  def to_boolean(value)
+    value == 1
+  end
+end
+
+BulkImport::Generic.new(ARGV.first).start