2019-05-02 18:17:27 -04:00
|
|
|
# frozen_string_literal: true
|
|
|
|
|
2018-12-09 23:58:07 -05:00
|
|
|
if ARGV.include?('bbcode-to-md')
|
|
|
|
# Replace (most) bbcode with markdown before creating posts.
|
|
|
|
# This will dramatically clean up the final posts in Discourse.
|
|
|
|
#
|
|
|
|
# In a temp dir:
|
|
|
|
#
|
|
|
|
# git clone https://github.com/nlalonde/ruby-bbcode-to-md.git
|
|
|
|
# cd ruby-bbcode-to-md
|
|
|
|
# gem build ruby-bbcode-to-md.gemspec
|
|
|
|
# gem install ruby-bbcode-to-md-*.gem
|
|
|
|
require 'ruby-bbcode-to-md'
|
|
|
|
end
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
require "pg"
|
|
|
|
require "set"
|
|
|
|
require "redcarpet"
|
2018-02-02 16:28:34 -05:00
|
|
|
require "htmlentities"
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
puts "Loading application..."
|
|
|
|
require_relative "../../config/environment"
|
2018-02-02 16:28:34 -05:00
|
|
|
require_relative '../import_scripts/base/uploader'
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
module BulkImport; end
|
|
|
|
|
|
|
|
class BulkImport::Base
|
|
|
|
|
2020-04-30 02:48:34 -04:00
|
|
|
NOW ||= "now()"
|
2017-04-24 16:57:15 -04:00
|
|
|
PRIVATE_OFFSET ||= 2**30
|
|
|
|
|
2019-12-09 19:48:27 -05:00
|
|
|
# rubocop:disable Layout/HashAlignment
|
2018-12-04 04:48:16 -05:00
|
|
|
|
2018-02-02 16:28:34 -05:00
|
|
|
CHARSET_MAP = {
|
|
|
|
"armscii8" => nil,
|
|
|
|
"ascii" => Encoding::US_ASCII,
|
|
|
|
"big5" => Encoding::Big5,
|
|
|
|
"binary" => Encoding::ASCII_8BIT,
|
|
|
|
"cp1250" => Encoding::Windows_1250,
|
|
|
|
"cp1251" => Encoding::Windows_1251,
|
|
|
|
"cp1256" => Encoding::Windows_1256,
|
|
|
|
"cp1257" => Encoding::Windows_1257,
|
|
|
|
"cp850" => Encoding::CP850,
|
|
|
|
"cp852" => Encoding::CP852,
|
|
|
|
"cp866" => Encoding::IBM866,
|
|
|
|
"cp932" => Encoding::Windows_31J,
|
|
|
|
"dec8" => nil,
|
|
|
|
"eucjpms" => Encoding::EucJP_ms,
|
|
|
|
"euckr" => Encoding::EUC_KR,
|
|
|
|
"gb2312" => Encoding::EUC_CN,
|
|
|
|
"gbk" => Encoding::GBK,
|
|
|
|
"geostd8" => nil,
|
|
|
|
"greek" => Encoding::ISO_8859_7,
|
|
|
|
"hebrew" => Encoding::ISO_8859_8,
|
|
|
|
"hp8" => nil,
|
|
|
|
"keybcs2" => nil,
|
|
|
|
"koi8r" => Encoding::KOI8_R,
|
|
|
|
"koi8u" => Encoding::KOI8_U,
|
|
|
|
"latin1" => Encoding::ISO_8859_1,
|
|
|
|
"latin2" => Encoding::ISO_8859_2,
|
|
|
|
"latin5" => Encoding::ISO_8859_9,
|
|
|
|
"latin7" => Encoding::ISO_8859_13,
|
|
|
|
"macce" => Encoding::MacCentEuro,
|
|
|
|
"macroman" => Encoding::MacRoman,
|
|
|
|
"sjis" => Encoding::SHIFT_JIS,
|
|
|
|
"swe7" => nil,
|
|
|
|
"tis620" => Encoding::TIS_620,
|
|
|
|
"ucs2" => Encoding::UTF_16BE,
|
|
|
|
"ujis" => Encoding::EucJP_ms,
|
|
|
|
"utf8" => Encoding::UTF_8,
|
|
|
|
}
|
|
|
|
|
2019-12-09 19:48:27 -05:00
|
|
|
# rubocop:enable Layout/HashAlignment
|
2018-12-04 04:48:16 -05:00
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
def initialize
|
2018-02-02 16:28:34 -05:00
|
|
|
charset = ENV["DB_CHARSET"] || "utf8"
|
2017-04-24 16:57:15 -04:00
|
|
|
db = ActiveRecord::Base.connection_config
|
|
|
|
@encoder = PG::TextEncoder::CopyRow.new
|
2020-07-14 01:58:27 -04:00
|
|
|
@raw_connection = PG.connect(dbname: db[:database], port: db[:port])
|
2018-02-02 16:28:34 -05:00
|
|
|
@uploader = ImportScripts::Uploader.new
|
|
|
|
@html_entities = HTMLEntities.new
|
|
|
|
@encoding = CHARSET_MAP[charset]
|
2018-12-09 23:58:07 -05:00
|
|
|
@bbcode_to_md = true if use_bbcode_to_md?
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
@markdown = Redcarpet::Markdown.new(
|
2017-09-04 05:04:54 -04:00
|
|
|
Redcarpet::Render::HTML.new(hard_wrap: true),
|
|
|
|
no_intra_emphasis: true,
|
2017-04-24 16:57:15 -04:00
|
|
|
fenced_code_blocks: true,
|
|
|
|
autolink: true
|
|
|
|
)
|
|
|
|
end
|
|
|
|
|
|
|
|
def run
|
|
|
|
puts "Starting..."
|
2018-02-02 16:28:34 -05:00
|
|
|
Rails.logger.level = 3 # :error, so that we don't create log files that are many GB
|
2017-04-24 16:57:15 -04:00
|
|
|
preload_i18n
|
|
|
|
fix_highest_post_numbers
|
|
|
|
load_imported_ids
|
|
|
|
load_indexes
|
|
|
|
execute
|
|
|
|
fix_primary_keys
|
2018-11-21 10:28:35 -05:00
|
|
|
puts "Done! Now run the 'import:ensure_consistency' rake task."
|
2017-04-24 16:57:15 -04:00
|
|
|
end
|
|
|
|
|
|
|
|
def preload_i18n
|
|
|
|
puts "Preloading I18n..."
|
2019-05-15 17:43:00 -04:00
|
|
|
I18n.locale = ENV.fetch("LOCALE") { SiteSettings::DefaultsProvider::DEFAULT_LOCALE }.to_sym
|
2017-04-24 16:57:15 -04:00
|
|
|
I18n.t("test")
|
|
|
|
ActiveSupport::Inflector.transliterate("test")
|
|
|
|
end
|
|
|
|
|
|
|
|
def fix_highest_post_numbers
|
|
|
|
puts "Fixing highest post numbers..."
|
|
|
|
@raw_connection.exec <<-SQL
|
|
|
|
WITH X AS (
|
|
|
|
SELECT topic_id
|
|
|
|
, COALESCE(MAX(post_number), 0) max_post_number
|
|
|
|
FROM posts
|
|
|
|
WHERE deleted_at IS NULL
|
|
|
|
GROUP BY topic_id
|
|
|
|
)
|
|
|
|
UPDATE topics
|
|
|
|
SET highest_post_number = X.max_post_number
|
|
|
|
FROM X
|
|
|
|
WHERE id = X.topic_id
|
|
|
|
AND highest_post_number <> X.max_post_number
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
2020-06-16 13:55:08 -04:00
|
|
|
def imported_ids(name)
|
|
|
|
map = []
|
|
|
|
ids = []
|
|
|
|
|
|
|
|
@raw_connection.send_query("SELECT value, #{name}_id FROM #{name}_custom_fields WHERE name = 'import_id'")
|
|
|
|
@raw_connection.set_single_row_mode
|
|
|
|
|
|
|
|
@raw_connection.get_result.stream_each do |row|
|
|
|
|
id = row["value"].to_i
|
|
|
|
ids << id
|
2020-06-18 11:42:14 -04:00
|
|
|
map[id] = row["#{name}_id"].to_i
|
2020-06-16 13:55:08 -04:00
|
|
|
end
|
|
|
|
|
|
|
|
@raw_connection.get_result
|
|
|
|
|
|
|
|
[map, ids]
|
|
|
|
end
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
def load_imported_ids
|
|
|
|
puts "Loading imported group ids..."
|
2020-06-16 13:55:08 -04:00
|
|
|
@groups, imported_group_ids = imported_ids("group")
|
|
|
|
@last_imported_group_id = imported_group_ids.max || -1
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
puts "Loading imported user ids..."
|
2020-06-16 13:55:08 -04:00
|
|
|
@users, imported_user_ids = imported_ids("user")
|
|
|
|
@last_imported_user_id = imported_user_ids.max || -1
|
2021-09-28 18:20:06 -04:00
|
|
|
@pre_existing_user_ids = Set.new
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
puts "Loading imported category ids..."
|
2020-06-16 13:55:08 -04:00
|
|
|
@categories, imported_category_ids = imported_ids("category")
|
|
|
|
@last_imported_category_id = imported_category_ids.max || -1
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
puts "Loading imported topic ids..."
|
2020-06-16 13:55:08 -04:00
|
|
|
@topics, imported_topic_ids = imported_ids("topic")
|
2017-04-24 16:57:15 -04:00
|
|
|
@last_imported_topic_id = imported_topic_ids.select { |id| id < PRIVATE_OFFSET }.max || -1
|
|
|
|
@last_imported_private_topic_id = imported_topic_ids.select { |id| id > PRIVATE_OFFSET }.max || (PRIVATE_OFFSET - 1)
|
|
|
|
|
|
|
|
puts "Loading imported post ids..."
|
2020-06-16 13:55:08 -04:00
|
|
|
@posts, imported_post_ids = imported_ids("post")
|
2017-04-24 16:57:15 -04:00
|
|
|
@last_imported_post_id = imported_post_ids.select { |id| id < PRIVATE_OFFSET }.max || -1
|
|
|
|
@last_imported_private_post_id = imported_post_ids.select { |id| id > PRIVATE_OFFSET }.max || (PRIVATE_OFFSET - 1)
|
|
|
|
end
|
|
|
|
|
2018-03-09 14:26:18 -05:00
|
|
|
def last_id(klass)
|
|
|
|
# the first record created will have id of this value + 1
|
|
|
|
[klass.unscoped.maximum(:id) || 0, 0].max
|
|
|
|
end
|
|
|
|
|
2020-06-18 10:32:27 -04:00
|
|
|
def load_values(name, column, size)
|
|
|
|
map = Array.new(size)
|
|
|
|
|
|
|
|
@raw_connection.send_query("SELECT id, #{column} FROM #{name}")
|
|
|
|
@raw_connection.set_single_row_mode
|
|
|
|
|
|
|
|
@raw_connection.get_result.stream_each do |row|
|
2020-06-18 11:42:14 -04:00
|
|
|
map[row["id"].to_i] = row[column].to_i
|
2020-06-18 10:32:27 -04:00
|
|
|
end
|
|
|
|
|
|
|
|
@raw_connection.get_result
|
|
|
|
|
|
|
|
map
|
|
|
|
end
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
def load_indexes
|
|
|
|
puts "Loading groups indexes..."
|
2020-06-18 10:32:27 -04:00
|
|
|
@last_group_id = last_id(Group)
|
2017-04-24 16:57:15 -04:00
|
|
|
@group_names = Group.unscoped.pluck(:name).map(&:downcase).to_set
|
|
|
|
|
|
|
|
puts "Loading users indexes..."
|
2018-03-09 14:26:18 -05:00
|
|
|
@last_user_id = last_id(User)
|
|
|
|
@last_user_email_id = last_id(UserEmail)
|
2021-09-28 18:20:06 -04:00
|
|
|
@emails = User.unscoped.joins(:user_emails).pluck(:"user_emails.email", :"user_emails.user_id").to_h
|
2017-04-24 16:57:15 -04:00
|
|
|
@usernames_lower = User.unscoped.pluck(:username_lower).to_set
|
|
|
|
@mapped_usernames = UserCustomField.joins(:user).where(name: "import_username").pluck("user_custom_fields.value", "users.username").to_h
|
|
|
|
|
|
|
|
puts "Loading categories indexes..."
|
2018-03-09 14:26:18 -05:00
|
|
|
@last_category_id = last_id(Category)
|
2017-04-24 16:57:15 -04:00
|
|
|
@category_names = Category.unscoped.pluck(:parent_category_id, :name).map { |pci, name| "#{pci}-#{name}" }.to_set
|
|
|
|
|
|
|
|
puts "Loading topics indexes..."
|
2018-03-09 14:26:18 -05:00
|
|
|
@last_topic_id = last_id(Topic)
|
2020-06-18 10:32:27 -04:00
|
|
|
@highest_post_number_by_topic_id = load_values("topics", "highest_post_number", @last_topic_id)
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
puts "Loading posts indexes..."
|
2018-03-09 14:26:18 -05:00
|
|
|
@last_post_id = last_id(Post)
|
2020-06-18 10:32:27 -04:00
|
|
|
@post_number_by_post_id = load_values("posts", "post_number", @last_post_id)
|
|
|
|
@topic_id_by_post_id = load_values("posts", "topic_id", @last_post_id)
|
2017-08-01 04:01:45 -04:00
|
|
|
|
|
|
|
puts "Loading post actions indexes..."
|
2018-03-09 14:26:18 -05:00
|
|
|
@last_post_action_id = last_id(PostAction)
|
2017-04-24 16:57:15 -04:00
|
|
|
end
|
|
|
|
|
2018-12-09 23:58:07 -05:00
|
|
|
def use_bbcode_to_md?
|
|
|
|
ARGV.include?("bbcode-to-md")
|
|
|
|
end
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
def execute
|
|
|
|
raise NotImplementedError
|
|
|
|
end
|
|
|
|
|
|
|
|
def fix_primary_keys
|
|
|
|
puts "Updating primary key sequences..."
|
2018-01-19 15:01:00 -05:00
|
|
|
@raw_connection.exec("SELECT setval('#{Group.sequence_name}', #{@last_group_id})") if @last_group_id > 0
|
|
|
|
@raw_connection.exec("SELECT setval('#{User.sequence_name}', #{@last_user_id})") if @last_user_id > 0
|
|
|
|
@raw_connection.exec("SELECT setval('#{UserEmail.sequence_name}', #{@last_user_email_id})") if @last_user_email_id > 0
|
|
|
|
@raw_connection.exec("SELECT setval('#{Category.sequence_name}', #{@last_category_id})") if @last_category_id > 0
|
|
|
|
@raw_connection.exec("SELECT setval('#{Topic.sequence_name}', #{@last_topic_id})") if @last_topic_id > 0
|
|
|
|
@raw_connection.exec("SELECT setval('#{Post.sequence_name}', #{@last_post_id})") if @last_post_id > 0
|
|
|
|
@raw_connection.exec("SELECT setval('#{PostAction.sequence_name}', #{@last_post_action_id})") if @last_post_action_id > 0
|
2017-04-24 16:57:15 -04:00
|
|
|
end
|
|
|
|
|
2018-12-03 22:48:13 -05:00
|
|
|
def group_id_from_imported_id(id)
|
2020-06-16 13:55:08 -04:00
|
|
|
@groups[id.to_i]
|
2018-12-03 22:48:13 -05:00
|
|
|
end
|
2020-06-16 13:55:08 -04:00
|
|
|
|
2018-12-03 22:48:13 -05:00
|
|
|
def user_id_from_imported_id(id)
|
2020-06-16 13:55:08 -04:00
|
|
|
@users[id.to_i]
|
2018-12-03 22:48:13 -05:00
|
|
|
end
|
2020-06-16 13:55:08 -04:00
|
|
|
|
2018-12-03 22:48:13 -05:00
|
|
|
def category_id_from_imported_id(id)
|
2020-06-16 13:55:08 -04:00
|
|
|
@categories[id.to_i]
|
2018-12-03 22:48:13 -05:00
|
|
|
end
|
2020-06-16 13:55:08 -04:00
|
|
|
|
2018-12-03 22:48:13 -05:00
|
|
|
def topic_id_from_imported_id(id)
|
2020-06-16 13:55:08 -04:00
|
|
|
@topics[id.to_i]
|
2018-12-03 22:48:13 -05:00
|
|
|
end
|
2020-06-16 13:55:08 -04:00
|
|
|
|
2018-12-03 22:48:13 -05:00
|
|
|
def post_id_from_imported_id(id)
|
2020-06-16 13:55:08 -04:00
|
|
|
@posts[id.to_i]
|
2018-12-03 22:48:13 -05:00
|
|
|
end
|
2017-04-24 16:57:15 -04:00
|
|
|
|
2018-12-03 22:48:13 -05:00
|
|
|
def post_number_from_imported_id(id)
|
2020-06-19 11:53:06 -04:00
|
|
|
post_id = post_id_from_imported_id(id)
|
|
|
|
post_id && @post_number_by_post_id[post_id]
|
2018-12-03 22:48:13 -05:00
|
|
|
end
|
2020-06-19 11:53:06 -04:00
|
|
|
|
2018-12-03 22:48:13 -05:00
|
|
|
def topic_id_from_imported_post_id(id)
|
2020-06-19 11:53:06 -04:00
|
|
|
post_id = post_id_from_imported_id(id)
|
|
|
|
post_id && @topic_id_by_post_id[post_id]
|
2018-12-03 22:48:13 -05:00
|
|
|
end
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
GROUP_COLUMNS ||= %i{
|
|
|
|
id name title bio_raw bio_cooked created_at updated_at
|
|
|
|
}
|
|
|
|
|
|
|
|
USER_COLUMNS ||= %i{
|
2017-07-24 08:22:00 -04:00
|
|
|
id username username_lower name active trust_level admin moderator
|
2017-04-24 16:57:15 -04:00
|
|
|
date_of_birth ip_address registration_ip_address primary_group_id
|
|
|
|
suspended_at suspended_till last_emailed_at created_at updated_at
|
|
|
|
}
|
|
|
|
|
2017-07-24 08:49:22 -04:00
|
|
|
USER_EMAIL_COLUMNS ||= %i{
|
|
|
|
id user_id email primary created_at updated_at
|
|
|
|
}
|
|
|
|
|
|
|
|
USER_STAT_COLUMNS ||= %i{
|
|
|
|
user_id topics_entered time_read days_visited posts_read_count
|
2020-07-14 01:58:27 -04:00
|
|
|
likes_given likes_received new_since read_faq
|
2017-07-24 08:49:22 -04:00
|
|
|
first_post_created_at post_count topic_count bounce_score
|
2020-12-13 18:58:14 -05:00
|
|
|
reset_bounce_score_after digest_attempted_at
|
2017-07-24 08:49:22 -04:00
|
|
|
}
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
USER_PROFILE_COLUMNS ||= %i{
|
|
|
|
user_id location website bio_raw bio_cooked views
|
|
|
|
}
|
|
|
|
|
|
|
|
GROUP_USER_COLUMNS ||= %i{
|
|
|
|
group_id user_id created_at updated_at
|
|
|
|
}
|
|
|
|
|
|
|
|
CATEGORY_COLUMNS ||= %i{
|
|
|
|
id name name_lower slug user_id description position parent_category_id
|
|
|
|
created_at updated_at
|
|
|
|
}
|
|
|
|
|
|
|
|
TOPIC_COLUMNS ||= %i{
|
|
|
|
id archetype title fancy_title slug user_id last_post_user_id category_id
|
|
|
|
visible closed pinned_at views created_at bumped_at updated_at
|
|
|
|
}
|
|
|
|
|
|
|
|
POST_COLUMNS ||= %i{
|
|
|
|
id user_id last_editor_id topic_id post_number sort_order reply_to_post_number
|
2017-07-31 04:56:57 -04:00
|
|
|
like_count raw cooked hidden word_count created_at last_version_at updated_at
|
2017-04-24 16:57:15 -04:00
|
|
|
}
|
|
|
|
|
2017-08-01 04:01:45 -04:00
|
|
|
POST_ACTION_COLUMNS ||= %i{
|
|
|
|
id post_id user_id post_action_type_id deleted_at created_at updated_at
|
|
|
|
deleted_by_id related_post_id staff_took_action deferred_by_id targets_topic
|
|
|
|
agreed_at agreed_by_id deferred_at disagreed_at disagreed_by_id
|
|
|
|
}
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
TOPIC_ALLOWED_USER_COLUMNS ||= %i{
|
|
|
|
topic_id user_id created_at updated_at
|
|
|
|
}
|
|
|
|
|
2018-02-02 16:28:34 -05:00
|
|
|
TOPIC_TAG_COLUMNS ||= %i{
|
|
|
|
topic_id tag_id created_at updated_at
|
|
|
|
}
|
|
|
|
|
2018-12-03 22:48:13 -05:00
|
|
|
def create_groups(rows, &block)
|
|
|
|
create_records(rows, "group", GROUP_COLUMNS, &block)
|
|
|
|
end
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
def create_users(rows, &block)
|
|
|
|
@imported_usernames = {}
|
|
|
|
|
|
|
|
create_records(rows, "user", USER_COLUMNS, &block)
|
|
|
|
|
|
|
|
create_custom_fields("user", "username", @imported_usernames.keys) do |username|
|
|
|
|
{
|
|
|
|
record_id: @imported_usernames[username],
|
|
|
|
value: username,
|
|
|
|
}
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2018-12-03 22:48:13 -05:00
|
|
|
def create_user_emails(rows, &block)
|
|
|
|
create_records(rows, "user_email", USER_EMAIL_COLUMNS, &block)
|
|
|
|
end
|
|
|
|
def create_user_stats(rows, &block)
|
|
|
|
create_records(rows, "user_stat", USER_STAT_COLUMNS, &block)
|
|
|
|
end
|
|
|
|
def create_user_profiles(rows, &block)
|
|
|
|
create_records(rows, "user_profile", USER_PROFILE_COLUMNS, &block)
|
|
|
|
end
|
|
|
|
def create_group_users(rows, &block)
|
|
|
|
create_records(rows, "group_user", GROUP_USER_COLUMNS, &block)
|
|
|
|
end
|
2019-04-11 02:36:37 -04:00
|
|
|
def create_categories(rows, &block)
|
|
|
|
create_records(rows, "category", CATEGORY_COLUMNS, &block)
|
|
|
|
end
|
2018-12-03 22:48:13 -05:00
|
|
|
def create_topics(rows, &block)
|
|
|
|
create_records(rows, "topic", TOPIC_COLUMNS, &block)
|
|
|
|
end
|
|
|
|
def create_posts(rows, &block)
|
|
|
|
create_records(rows, "post", POST_COLUMNS, &block)
|
|
|
|
end
|
|
|
|
def create_post_actions(rows, &block)
|
|
|
|
create_records(rows, "post_action", POST_ACTION_COLUMNS, &block)
|
|
|
|
end
|
|
|
|
def create_topic_allowed_users(rows, &block)
|
|
|
|
create_records(rows, "topic_allowed_user", TOPIC_ALLOWED_USER_COLUMNS, &block)
|
|
|
|
end
|
|
|
|
def create_topic_tags(rows, &block)
|
|
|
|
create_records(rows, "topic_tag", TOPIC_TAG_COLUMNS, &block)
|
|
|
|
end
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
def process_group(group)
|
2020-06-16 13:55:08 -04:00
|
|
|
@groups[group[:imported_id].to_i] = group[:id] = @last_group_id += 1
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
group[:name] = fix_name(group[:name])
|
|
|
|
|
|
|
|
unless @group_names.add?(group[:name].downcase)
|
|
|
|
group_name = group[:name] + "_1"
|
|
|
|
group_name.next! until @group_names.add?(group_name.downcase)
|
|
|
|
group[:name] = group_name
|
|
|
|
end
|
|
|
|
|
2017-08-03 11:43:02 -04:00
|
|
|
group[:title] = group[:title].scrub.strip.presence if group[:title].present?
|
|
|
|
group[:bio_raw] = group[:bio_raw].scrub.strip.presence if group[:bio_raw].present?
|
2017-04-24 16:57:15 -04:00
|
|
|
group[:bio_cooked] = pre_cook(group[:bio_raw]) if group[:bio_raw].present?
|
|
|
|
group[:created_at] ||= NOW
|
|
|
|
group[:updated_at] ||= group[:created_at]
|
|
|
|
group
|
|
|
|
end
|
|
|
|
|
|
|
|
def process_user(user)
|
2021-09-28 18:20:06 -04:00
|
|
|
if user[:email].present?
|
|
|
|
user[:email].downcase!
|
|
|
|
|
|
|
|
if existing_user_id = @emails[user[:email]]
|
|
|
|
@pre_existing_user_ids << existing_user_id
|
|
|
|
@users[user[:imported_id].to_i] = existing_user_id
|
|
|
|
user[:skip] = true
|
|
|
|
return user
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2020-06-16 13:55:08 -04:00
|
|
|
@users[user[:imported_id].to_i] = user[:id] = @last_user_id += 1
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
imported_username = user[:username].dup
|
|
|
|
|
|
|
|
user[:username] = fix_name(user[:username]).presence || random_username
|
|
|
|
|
|
|
|
if user[:username] != imported_username
|
|
|
|
@imported_usernames[imported_username] = user[:id]
|
|
|
|
@mapped_usernames[imported_username] = user[:username]
|
|
|
|
end
|
|
|
|
|
|
|
|
# unique username_lower
|
|
|
|
unless @usernames_lower.add?(user[:username].downcase)
|
|
|
|
username = user[:username] + "_1"
|
|
|
|
username.next! until @usernames_lower.add?(username.downcase)
|
|
|
|
user[:username] = username
|
|
|
|
end
|
|
|
|
|
|
|
|
user[:username_lower] = user[:username].downcase
|
|
|
|
user[:trust_level] ||= TrustLevel[1]
|
|
|
|
user[:active] = true unless user.has_key?(:active)
|
|
|
|
user[:admin] ||= false
|
|
|
|
user[:moderator] ||= false
|
|
|
|
user[:last_emailed_at] ||= NOW
|
|
|
|
user[:created_at] ||= NOW
|
|
|
|
user[:updated_at] ||= user[:created_at]
|
|
|
|
user
|
|
|
|
end
|
|
|
|
|
2017-07-24 08:49:22 -04:00
|
|
|
def process_user_email(user_email)
|
2021-09-28 18:20:06 -04:00
|
|
|
user_id = @users[user_email[:imported_user_id].to_i]
|
|
|
|
return { skip: true } if @pre_existing_user_ids.include?(user_id)
|
|
|
|
|
2018-12-03 22:48:13 -05:00
|
|
|
user_email[:id] = @last_user_email_id += 1
|
2021-09-28 18:20:06 -04:00
|
|
|
user_email[:user_id] = user_id
|
2017-07-24 08:49:22 -04:00
|
|
|
user_email[:primary] = true
|
|
|
|
user_email[:created_at] ||= NOW
|
|
|
|
user_email[:updated_at] ||= user_email[:created_at]
|
2021-09-28 18:20:06 -04:00
|
|
|
|
2017-07-24 08:49:22 -04:00
|
|
|
user_email[:email] ||= random_email
|
|
|
|
user_email[:email].downcase!
|
2021-09-28 18:20:06 -04:00
|
|
|
# unique email
|
|
|
|
user_email[:email] = random_email until user_email[:email] =~ EmailValidator.email_regex && !@emails.has_key?(user_email[:email])
|
|
|
|
|
2017-07-24 08:49:22 -04:00
|
|
|
user_email
|
|
|
|
end
|
|
|
|
|
|
|
|
def process_user_stat(user_stat)
|
2021-09-28 18:20:06 -04:00
|
|
|
user_id = @users[user_stat[:imported_user_id].to_i]
|
|
|
|
return { skip: true } if @pre_existing_user_ids.include?(user_id)
|
|
|
|
|
|
|
|
user_stat[:user_id] = user_id
|
2017-07-24 08:49:22 -04:00
|
|
|
user_stat[:topics_entered] ||= 0
|
|
|
|
user_stat[:time_read] ||= 0
|
|
|
|
user_stat[:days_visited] ||= 0
|
|
|
|
user_stat[:posts_read_count] ||= 0
|
|
|
|
user_stat[:likes_given] ||= 0
|
|
|
|
user_stat[:likes_received] ||= 0
|
|
|
|
user_stat[:new_since] ||= NOW
|
|
|
|
user_stat[:post_count] ||= 0
|
|
|
|
user_stat[:topic_count] ||= 0
|
|
|
|
user_stat[:bounce_score] ||= 0
|
2020-12-13 18:58:14 -05:00
|
|
|
user_stat[:digest_attempted_at] ||= NOW
|
2017-07-24 08:49:22 -04:00
|
|
|
user_stat
|
|
|
|
end
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
def process_user_profile(user_profile)
|
2021-09-28 18:20:06 -04:00
|
|
|
return { skip: true } if @pre_existing_user_ids.include?(user_profile[:user_id])
|
|
|
|
|
2020-12-13 18:58:14 -05:00
|
|
|
user_profile[:bio_raw] = (user_profile[:bio_raw].presence || "").scrub.strip.presence
|
2017-04-24 16:57:15 -04:00
|
|
|
user_profile[:bio_cooked] = pre_cook(user_profile[:bio_raw]) if user_profile[:bio_raw].present?
|
2017-08-03 11:43:02 -04:00
|
|
|
user_profile[:views] ||= 0
|
2017-04-24 16:57:15 -04:00
|
|
|
user_profile
|
|
|
|
end
|
|
|
|
|
|
|
|
def process_group_user(group_user)
|
|
|
|
group_user[:created_at] = NOW
|
|
|
|
group_user[:updated_at] = NOW
|
|
|
|
group_user
|
|
|
|
end
|
|
|
|
|
|
|
|
def process_category(category)
|
2017-07-31 04:56:57 -04:00
|
|
|
category[:id] ||= @last_category_id += 1
|
2020-06-16 13:55:08 -04:00
|
|
|
@categories[category[:imported_id].to_i] ||= category[:id]
|
2017-04-24 16:57:15 -04:00
|
|
|
category[:name] = category[:name][0...50].scrub.strip
|
|
|
|
# TODO: unique name
|
|
|
|
category[:name_lower] = category[:name].downcase
|
|
|
|
category[:slug] ||= Slug.ascii_generator(category[:name_lower])
|
|
|
|
category[:description] = (category[:description] || "").scrub.strip.presence
|
|
|
|
category[:user_id] ||= Discourse::SYSTEM_USER_ID
|
|
|
|
category[:created_at] ||= NOW
|
|
|
|
category[:updated_at] ||= category[:created_at]
|
|
|
|
category
|
|
|
|
end
|
|
|
|
|
|
|
|
def process_topic(topic)
|
2020-06-16 13:55:08 -04:00
|
|
|
@topics[topic[:imported_id].to_i] = topic[:id] = @last_topic_id += 1
|
2017-04-24 16:57:15 -04:00
|
|
|
topic[:archetype] ||= Archetype.default
|
|
|
|
topic[:title] = topic[:title][0...255].scrub.strip
|
|
|
|
topic[:fancy_title] ||= pre_fancy(topic[:title])
|
|
|
|
topic[:slug] ||= Slug.ascii_generator(topic[:title])
|
|
|
|
topic[:user_id] ||= Discourse::SYSTEM_USER_ID
|
|
|
|
topic[:last_post_user_id] ||= topic[:user_id]
|
|
|
|
topic[:category_id] ||= -1 if topic[:archetype] != Archetype.private_message
|
|
|
|
topic[:visible] = true unless topic.has_key?(:visible)
|
|
|
|
topic[:closed] ||= false
|
|
|
|
topic[:views] ||= 0
|
|
|
|
topic[:created_at] ||= NOW
|
2020-12-13 18:58:14 -05:00
|
|
|
topic[:bumped_at] ||= topic[:created_at]
|
2017-04-24 16:57:15 -04:00
|
|
|
topic[:updated_at] ||= topic[:created_at]
|
|
|
|
topic
|
|
|
|
end
|
|
|
|
|
|
|
|
def process_post(post)
|
2020-06-16 13:55:08 -04:00
|
|
|
@posts[post[:imported_id].to_i] = post[:id] = @last_post_id += 1
|
2017-04-24 16:57:15 -04:00
|
|
|
post[:user_id] ||= Discourse::SYSTEM_USER_ID
|
|
|
|
post[:last_editor_id] = post[:user_id]
|
|
|
|
@highest_post_number_by_topic_id[post[:topic_id]] ||= 0
|
|
|
|
post[:post_number] = @highest_post_number_by_topic_id[post[:topic_id]] += 1
|
|
|
|
post[:sort_order] = post[:post_number]
|
|
|
|
@post_number_by_post_id[post[:id]] = post[:post_number]
|
|
|
|
@topic_id_by_post_id[post[:id]] = post[:topic_id]
|
|
|
|
post[:raw] = (post[:raw] || "").scrub.strip.presence || "<Empty imported post>"
|
|
|
|
post[:raw] = process_raw post[:raw]
|
2018-12-09 23:58:07 -05:00
|
|
|
if @bbcode_to_md
|
2018-12-11 09:39:46 -05:00
|
|
|
post[:raw] = post[:raw].bbcode_to_md(false, {}, :disable, :quote) rescue post[:raw]
|
2018-12-09 23:58:07 -05:00
|
|
|
end
|
2017-07-31 04:56:57 -04:00
|
|
|
post[:like_count] ||= 0
|
2017-04-24 16:57:15 -04:00
|
|
|
post[:cooked] = pre_cook post[:raw]
|
|
|
|
post[:hidden] ||= false
|
|
|
|
post[:word_count] = post[:raw].scan(/[[:word:]]+/).size
|
|
|
|
post[:created_at] ||= NOW
|
|
|
|
post[:last_version_at] = post[:created_at]
|
|
|
|
post[:updated_at] ||= post[:created_at]
|
|
|
|
post
|
|
|
|
end
|
|
|
|
|
2017-08-01 04:01:45 -04:00
|
|
|
def process_post_action(post_action)
|
|
|
|
post_action[:id] ||= @last_post_action_id += 1
|
|
|
|
post_action[:staff_took_action] ||= false
|
|
|
|
post_action[:targets_topic] ||= false
|
|
|
|
post_action[:created_at] ||= NOW
|
|
|
|
post_action[:updated_at] ||= post_action[:created_at]
|
|
|
|
post_action
|
|
|
|
end
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
def process_topic_allowed_user(topic_allowed_user)
|
|
|
|
topic_allowed_user[:created_at] = NOW
|
|
|
|
topic_allowed_user[:updated_at] = NOW
|
|
|
|
topic_allowed_user
|
|
|
|
end
|
|
|
|
|
2018-02-02 16:28:34 -05:00
|
|
|
def process_topic_tag(topic_tag)
|
|
|
|
topic_tag[:created_at] = NOW
|
|
|
|
topic_tag[:updated_at] = NOW
|
|
|
|
topic_tag
|
|
|
|
end
|
|
|
|
|
2020-07-14 01:58:27 -04:00
|
|
|
def process_raw(original_raw)
|
|
|
|
raw = original_raw.dup
|
2017-04-24 16:57:15 -04:00
|
|
|
# fix whitespaces
|
|
|
|
raw.gsub!(/(\\r)?\\n/, "\n")
|
|
|
|
raw.gsub!("\\t", "\t")
|
|
|
|
|
|
|
|
# [HTML]...[/HTML]
|
|
|
|
raw.gsub!(/\[HTML\]/i, "\n\n```html\n")
|
|
|
|
raw.gsub!(/\[\/HTML\]/i, "\n```\n\n")
|
|
|
|
|
|
|
|
# [PHP]...[/PHP]
|
|
|
|
raw.gsub!(/\[PHP\]/i, "\n\n```php\n")
|
|
|
|
raw.gsub!(/\[\/PHP\]/i, "\n```\n\n")
|
|
|
|
|
|
|
|
# [HIGHLIGHT="..."]
|
|
|
|
raw.gsub!(/\[HIGHLIGHT="?(\w+)"?\]/i) { "\n\n```#{$1.downcase}\n" }
|
|
|
|
|
|
|
|
# [CODE]...[/CODE]
|
|
|
|
# [HIGHLIGHT]...[/HIGHLIGHT]
|
|
|
|
raw.gsub!(/\[\/?CODE\]/i, "\n\n```\n\n")
|
|
|
|
raw.gsub!(/\[\/?HIGHLIGHT\]/i, "\n\n```\n\n")
|
|
|
|
|
|
|
|
# [SAMP]...[/SAMP]
|
|
|
|
raw.gsub!(/\[\/?SAMP\]/i, "`")
|
|
|
|
|
|
|
|
# replace all chevrons with HTML entities
|
|
|
|
# /!\ must be done /!\
|
|
|
|
# - AFTER the "code" processing
|
|
|
|
# - BEFORE the "quote" processing
|
|
|
|
raw.gsub!(/`([^`]+?)`/im) { "`" + $1.gsub("<", "\u2603") + "`" }
|
|
|
|
raw.gsub!("<", "<")
|
|
|
|
raw.gsub!("\u2603", "<")
|
|
|
|
|
|
|
|
raw.gsub!(/`([^`]+?)`/im) { "`" + $1.gsub(">", "\u2603") + "`" }
|
|
|
|
raw.gsub!(">", ">")
|
|
|
|
raw.gsub!("\u2603", ">")
|
|
|
|
|
|
|
|
raw.gsub!(/\[\/?I\]/i, "*")
|
|
|
|
raw.gsub!(/\[\/?B\]/i, "**")
|
|
|
|
raw.gsub!(/\[\/?U\]/i, "")
|
|
|
|
|
|
|
|
raw.gsub!(/\[\/?RED\]/i, "")
|
|
|
|
raw.gsub!(/\[\/?BLUE\]/i, "")
|
|
|
|
|
|
|
|
raw.gsub!(/\[AUTEUR\].+?\[\/AUTEUR\]/im, "")
|
|
|
|
raw.gsub!(/\[VOIRMSG\].+?\[\/VOIRMSG\]/im, "")
|
|
|
|
raw.gsub!(/\[PSEUDOID\].+?\[\/PSEUDOID\]/im, "")
|
|
|
|
|
|
|
|
# [IMG]...[/IMG]
|
|
|
|
raw.gsub!(/(?:\s*\[IMG\]\s*)+(.+?)(?:\s*\[\/IMG\]\s*)+/im) { "\n\n#{$1}\n\n" }
|
|
|
|
|
2018-02-02 16:28:34 -05:00
|
|
|
# [IMG=url]
|
|
|
|
raw.gsub!(/\[IMG=([^\]]*)\]/im) { "\n\n#{$1}\n\n" }
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
# [URL=...]...[/URL]
|
|
|
|
raw.gsub!(/\[URL="?(.+?)"?\](.+?)\[\/URL\]/im) { "[#{$2.strip}](#{$1})" }
|
|
|
|
|
|
|
|
# [URL]...[/URL]
|
|
|
|
# [MP3]...[/MP3]
|
2018-12-09 23:40:44 -05:00
|
|
|
# [EMAIL]...[/EMAIL]
|
|
|
|
# [LEFT]...[/LEFT]
|
2017-04-24 16:57:15 -04:00
|
|
|
raw.gsub!(/\[\/?URL\]/i, "")
|
|
|
|
raw.gsub!(/\[\/?MP3\]/i, "")
|
2018-12-09 23:40:44 -05:00
|
|
|
raw.gsub!(/\[\/?EMAIL\]/i, "")
|
|
|
|
raw.gsub!(/\[\/?LEFT\]/i, "")
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
# [FONT=blah] and [COLOR=blah]
|
2017-05-03 10:56:52 -04:00
|
|
|
raw.gsub!(/\[FONT=.*?\](.*?)\[\/FONT\]/im, "\\1")
|
|
|
|
raw.gsub!(/\[COLOR=.*?\](.*?)\[\/COLOR\]/im, "\\1")
|
2017-04-24 16:57:15 -04:00
|
|
|
|
2017-05-03 10:56:52 -04:00
|
|
|
raw.gsub!(/\[SIZE=.*?\](.*?)\[\/SIZE\]/im, "\\1")
|
|
|
|
raw.gsub!(/\[H=.*?\](.*?)\[\/H\]/im, "\\1")
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
# [CENTER]...[/CENTER]
|
2017-05-03 10:56:52 -04:00
|
|
|
raw.gsub!(/\[CENTER\](.*?)\[\/CENTER\]/im, "\\1")
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
# [INDENT]...[/INDENT]
|
2017-05-03 10:56:52 -04:00
|
|
|
raw.gsub!(/\[INDENT\](.*?)\[\/INDENT\]/im, "\\1")
|
|
|
|
raw.gsub!(/\[TABLE\](.*?)\[\/TABLE\]/im, "\\1")
|
|
|
|
raw.gsub!(/\[TR\](.*?)\[\/TR\]/im, "\\1")
|
|
|
|
raw.gsub!(/\[TD\](.*?)\[\/TD\]/im, "\\1")
|
|
|
|
raw.gsub!(/\[TD="?.*?"?\](.*?)\[\/TD\]/im, "\\1")
|
2017-04-24 16:57:15 -04:00
|
|
|
|
2021-04-30 12:03:33 -04:00
|
|
|
# [STRIKE]
|
|
|
|
raw.gsub!(/\[STRIKE\]/i, "<s>")
|
|
|
|
raw.gsub!(/\[\/STRIKE\]/i, "</s>")
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
# [QUOTE]...[/QUOTE]
|
2018-02-02 16:28:34 -05:00
|
|
|
raw.gsub!(/\[QUOTE="([^\]]+)"\]/i) { "[QUOTE=#{$1}]" }
|
|
|
|
|
|
|
|
# Nested Quotes
|
|
|
|
raw.gsub!(/(\[\/?QUOTE.*?\])/mi) { |q| "\n#{q}\n" }
|
2017-04-24 16:57:15 -04:00
|
|
|
|
2018-02-02 16:28:34 -05:00
|
|
|
# raw.gsub!(/\[QUOTE\](.+?)\[\/QUOTE\]/im) { |quote|
|
|
|
|
# quote.gsub!(/\[QUOTE\](.+?)\[\/QUOTE\]/im) { "\n#{$1}\n" }
|
|
|
|
# quote.gsub!(/\n(.+?)/) { "\n> #{$1}" }
|
|
|
|
# }
|
|
|
|
|
|
|
|
# [QUOTE=<username>;<postid>]
|
|
|
|
raw.gsub!(/\[QUOTE=([^;\]]+);(\d+)\]/i) do
|
|
|
|
imported_username, imported_postid = $1, $2
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
username = @mapped_usernames[imported_username] || imported_username
|
2020-06-19 11:53:06 -04:00
|
|
|
post_number = post_number_from_imported_id(imported_postid)
|
2021-04-30 12:03:33 -04:00
|
|
|
topic_id = topic_id_from_imported_post_id(imported_postid)
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
if post_number && topic_id
|
2018-02-02 16:28:34 -05:00
|
|
|
"\n[quote=\"#{username}, post:#{post_number}, topic:#{topic_id}\"]\n"
|
2017-04-24 16:57:15 -04:00
|
|
|
else
|
2018-02-02 16:28:34 -05:00
|
|
|
"\n[quote=\"#{username}\"]\n"
|
2017-04-24 16:57:15 -04:00
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
# [YOUTUBE]<id>[/YOUTUBE]
|
|
|
|
raw.gsub!(/\[YOUTUBE\](.+?)\[\/YOUTUBE\]/i) { "\nhttps://www.youtube.com/watch?v=#{$1}\n" }
|
|
|
|
raw.gsub!(/\[DAILYMOTION\](.+?)\[\/DAILYMOTION\]/i) { "\nhttps://www.dailymotion.com/video/#{$1}\n" }
|
|
|
|
|
|
|
|
# [VIDEO=youtube;<id>]...[/VIDEO]
|
|
|
|
raw.gsub!(/\[VIDEO=YOUTUBE;([^\]]+)\].*?\[\/VIDEO\]/i) { "\nhttps://www.youtube.com/watch?v=#{$1}\n" }
|
|
|
|
raw.gsub!(/\[VIDEO=DAILYMOTION;([^\]]+)\].*?\[\/VIDEO\]/i) { "\nhttps://www.dailymotion.com/video/#{$1}\n" }
|
|
|
|
|
|
|
|
# [SPOILER=Some hidden stuff]SPOILER HERE!![/SPOILER]
|
|
|
|
raw.gsub!(/\[SPOILER="?(.+?)"?\](.+?)\[\/SPOILER\]/im) { "\n#{$1}\n[spoiler]#{$2}[/spoiler]\n" }
|
|
|
|
|
2018-12-09 23:40:44 -05:00
|
|
|
# convert list tags to ul and list=1 tags to ol
|
|
|
|
# (basically, we're only missing list=a here...)
|
|
|
|
# (https://meta.discourse.org/t/phpbb-3-importer-old/17397)
|
|
|
|
raw.gsub!(/\[list\](.*?)\[\/list\]/im, '[ul]\1[/ul]')
|
2021-04-30 12:03:33 -04:00
|
|
|
raw.gsub!(/\[list=1\|?[^\]]*\](.*?)\[\/list\]/im, '[ol]\1[/ol]')
|
2018-12-09 23:40:44 -05:00
|
|
|
raw.gsub!(/\[list\](.*?)\[\/list:u\]/im, '[ul]\1[/ul]')
|
2021-04-30 12:03:33 -04:00
|
|
|
raw.gsub!(/\[list=1\|?[^\]]*\](.*?)\[\/list:o\]/im, '[ol]\1[/ol]')
|
2018-12-09 23:40:44 -05:00
|
|
|
# convert *-tags to li-tags so bbcode-to-md can do its magic on phpBB's lists:
|
|
|
|
raw.gsub!(/\[\*\]\n/, '')
|
|
|
|
raw.gsub!(/\[\*\](.*?)\[\/\*:m\]/, '[li]\1[/li]')
|
|
|
|
raw.gsub!(/\[\*\](.*?)\n/, '[li]\1[/li]')
|
|
|
|
raw.gsub!(/\[\*=1\]/, '')
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
raw
|
|
|
|
end
|
|
|
|
|
|
|
|
def create_records(rows, name, columns)
|
|
|
|
start = Time.now
|
|
|
|
imported_ids = []
|
|
|
|
process_method_name = "process_#{name}"
|
2017-07-24 08:49:22 -04:00
|
|
|
sql = "COPY #{name.pluralize} (#{columns.map { |c| "\"#{c}\"" }.join(",")}) FROM STDIN"
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
@raw_connection.copy_data(sql, @encoder) do
|
|
|
|
rows.each do |row|
|
2018-08-16 12:30:26 -04:00
|
|
|
begin
|
2020-06-19 11:53:06 -04:00
|
|
|
next unless mapped = yield(row)
|
2018-08-16 12:30:26 -04:00
|
|
|
processed = send(process_method_name, mapped)
|
|
|
|
imported_ids << mapped[:imported_id] unless mapped[:imported_id].nil?
|
|
|
|
imported_ids |= mapped[:imported_ids] unless mapped[:imported_ids].nil?
|
2021-09-28 18:20:06 -04:00
|
|
|
@raw_connection.put_copy_data columns.map { |c| processed[c] } unless processed[:skip]
|
2020-04-30 02:48:34 -04:00
|
|
|
print "\r%7d - %6d/sec" % [imported_ids.size, imported_ids.size.to_f / (Time.now - start)] if imported_ids.size % 5000 == 0
|
2018-08-16 12:30:26 -04:00
|
|
|
rescue => e
|
|
|
|
puts "\n"
|
2020-06-19 11:53:06 -04:00
|
|
|
puts "ERROR: #{e.message}"
|
2020-07-14 01:58:27 -04:00
|
|
|
puts e.backtrace.join("\n")
|
2018-08-16 12:30:26 -04:00
|
|
|
end
|
2017-04-24 16:57:15 -04:00
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
if imported_ids.size > 0
|
2020-04-30 02:48:34 -04:00
|
|
|
print "\r%7d - %6d/sec" % [imported_ids.size, imported_ids.size.to_f / (Time.now - start)]
|
2017-04-24 16:57:15 -04:00
|
|
|
puts
|
|
|
|
end
|
|
|
|
|
|
|
|
id_mapping_method_name = "#{name}_id_from_imported_id".freeze
|
|
|
|
return unless respond_to?(id_mapping_method_name)
|
|
|
|
create_custom_fields(name, "id", imported_ids) do |imported_id|
|
|
|
|
{
|
|
|
|
record_id: send(id_mapping_method_name, imported_id),
|
|
|
|
value: imported_id,
|
|
|
|
}
|
|
|
|
end
|
2017-08-03 11:43:02 -04:00
|
|
|
rescue => e
|
|
|
|
puts e.message
|
|
|
|
puts e.backtrace.join("\n")
|
2017-04-24 16:57:15 -04:00
|
|
|
end
|
|
|
|
|
|
|
|
def create_custom_fields(table, name, rows)
|
|
|
|
name = "import_#{name}"
|
|
|
|
sql = "COPY #{table}_custom_fields (#{table}_id, name, value, created_at, updated_at) FROM STDIN"
|
|
|
|
@raw_connection.copy_data(sql, @encoder) do
|
|
|
|
rows.each do |row|
|
2020-06-19 11:53:06 -04:00
|
|
|
next unless cf = yield(row)
|
2017-04-24 16:57:15 -04:00
|
|
|
@raw_connection.put_copy_data [cf[:record_id], name, cf[:value], NOW, NOW]
|
|
|
|
end
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2018-02-02 16:28:34 -05:00
|
|
|
def create_upload(user_id, path, source_filename)
|
|
|
|
@uploader.create_upload(user_id, path, source_filename)
|
|
|
|
end
|
|
|
|
|
2018-08-16 12:30:26 -04:00
|
|
|
def html_for_upload(upload, display_filename)
|
|
|
|
@uploader.html_for_upload(upload, display_filename)
|
|
|
|
end
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
def fix_name(name)
|
2017-07-31 15:34:11 -04:00
|
|
|
name.scrub! if name.valid_encoding? == false
|
2017-04-24 16:57:15 -04:00
|
|
|
return if name.blank?
|
|
|
|
name = ActiveSupport::Inflector.transliterate(name)
|
|
|
|
name.gsub!(/[^\w.-]+/, "_")
|
|
|
|
name.gsub!(/^\W+/, "")
|
|
|
|
name.gsub!(/[^A-Za-z0-9]+$/, "")
|
|
|
|
name.gsub!(/([-_.]{2,})/) { $1.first }
|
|
|
|
name.strip!
|
2021-04-30 12:03:33 -04:00
|
|
|
name.truncate(60)
|
2017-04-24 16:57:15 -04:00
|
|
|
name
|
|
|
|
end
|
|
|
|
|
|
|
|
def random_username
|
|
|
|
"Anonymous_#{SecureRandom.hex}"
|
|
|
|
end
|
|
|
|
|
|
|
|
def random_email
|
2021-04-30 12:03:33 -04:00
|
|
|
"#{SecureRandom.hex}@email.invalid"
|
2017-04-24 16:57:15 -04:00
|
|
|
end
|
|
|
|
|
|
|
|
def pre_cook(raw)
|
2017-07-31 04:56:57 -04:00
|
|
|
cooked = raw
|
|
|
|
|
|
|
|
# Convert YouTube URLs to lazyYT DOMs before being transformed into links
|
|
|
|
cooked.gsub!(/\nhttps\:\/\/www.youtube.com\/watch\?v=(\w+)\n/) do
|
|
|
|
video_id = $1
|
|
|
|
result = <<-HTML
|
|
|
|
<div class="lazyYT" data-youtube-id="#{video_id}" data-width="480" data-height="270" data-parameters="feature=oembed&wmode=opaque"></div>
|
|
|
|
HTML
|
|
|
|
result.strip
|
|
|
|
end
|
|
|
|
|
|
|
|
cooked = @markdown.render(cooked).scrub.strip
|
2017-04-24 16:57:15 -04:00
|
|
|
|
|
|
|
cooked.gsub!(/\[QUOTE="?([^,"]+)(?:, post:(\d+), topic:(\d+))?"?\](.+?)\[\/QUOTE\]/im) do
|
2017-09-04 05:04:54 -04:00
|
|
|
username, post_id, topic_id, quote = $1, $2, $3, $4
|
|
|
|
|
|
|
|
quote = quote.scrub.strip
|
|
|
|
quote.gsub!(/^(<br>\n?)+/, "")
|
|
|
|
quote.gsub!(/(<br>\n?)+$/, "")
|
2017-04-24 16:57:15 -04:00
|
|
|
|
2020-07-14 01:58:27 -04:00
|
|
|
user = User.find_by(username: username)
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
if post_id.present? && topic_id.present?
|
|
|
|
<<-HTML
|
|
|
|
<aside class="quote" data-post="#{post_id}" data-topic="#{topic_id}">
|
2020-07-14 01:58:27 -04:00
|
|
|
<div class="title">
|
|
|
|
<div class="quote-controls"></div>
|
|
|
|
#{user ? user_avatar(user) : username}:
|
|
|
|
</div>
|
2017-04-24 16:57:15 -04:00
|
|
|
<blockquote>#{quote}</blockquote>
|
|
|
|
</aside>
|
|
|
|
HTML
|
|
|
|
else
|
|
|
|
<<-HTML
|
2020-07-14 01:58:27 -04:00
|
|
|
<aside class="quote no-group" data-username="#{username}">
|
|
|
|
<div class="title">
|
|
|
|
<div class="quote-controls"></div>
|
|
|
|
#{user ? user_avatar(user) : username}:
|
|
|
|
</div>
|
2017-04-24 16:57:15 -04:00
|
|
|
<blockquote>#{quote}</blockquote>
|
|
|
|
</aside>
|
|
|
|
HTML
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
cooked.scrub.strip
|
|
|
|
end
|
|
|
|
|
2020-07-14 01:58:27 -04:00
|
|
|
def user_avatar(user)
|
|
|
|
url = user.avatar_template.gsub("{size}", "45")
|
|
|
|
"<img alt=\"\" width=\"20\" height=\"20\" src=\"#{url}\" class=\"avatar\"> #{user.username}"
|
|
|
|
end
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
def pre_fancy(title)
|
|
|
|
Redcarpet::Render::SmartyPants.render(ERB::Util.html_escape(title)).scrub.strip
|
|
|
|
end
|
|
|
|
|
2018-02-02 16:28:34 -05:00
|
|
|
def normalize_text(text)
|
|
|
|
return nil unless text.present?
|
|
|
|
@html_entities.decode(normalize_charset(text.presence || "").scrub)
|
|
|
|
end
|
|
|
|
|
|
|
|
def normalize_charset(text)
|
|
|
|
return text if @encoding == Encoding::UTF_8
|
2019-11-14 15:10:51 -05:00
|
|
|
text && text.encode(@encoding).force_encoding(Encoding::UTF_8)
|
2018-02-02 16:28:34 -05:00
|
|
|
end
|
|
|
|
|
2017-04-24 16:57:15 -04:00
|
|
|
end
|