Discource-C
/
discourse
mirror of https://github.com/discourse/discourse.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

FEATURE: vBulletin importer

This commit is contained in:
Régis Hanol 2014-08-18 13:04:08 +02:00
parent 625c3e2961
commit 66aaa9329e
4 changed files with 639 additions and 31 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
app/models/concerns/has_custom_fields.rb
View File

@ -1,6 +1,8 @@
module HasCustomFields
extend ActiveSupport::Concern
module Helpers
def self.append_field(target, key, value, types)
if target.has_key?(key)
target[key] = [target[key]] if !target[key].is_a? Array
@ -48,7 +50,6 @@ module HasCustomFields
HasCustomFields::Helpers.append_field(target,key,value,@custom_field_types)
end
def self.register_custom_field_type(name, type)
@custom_field_types ||= {}
@custom_field_types[name] = type

1
app/models/group.rb
View File

@ -269,6 +269,7 @@ class Group < ActiveRecord::Base
def add(user)
self.users.push(user)
end
protected
def name_format_validator

46
script/import_scripts/base.rb
View File

@ -44,8 +44,12 @@ class ImportScripts::Base
@existing_posts[import_id] = post_id
end
Post.pluck(:id, :topic_id, :post_number).each do |post_id,t,n|
@topic_lookup[post_id] = {topic_id: t, post_number: n}
Post.joins(:topic).select("posts.id, posts.topic_id, posts.post_number, topics.slug").each do |post|
@topic_lookup[post.id] = {
topic_id: post.topic_id,
post_number: post.post_number,
url: post.url,
}
end
end
@ -202,21 +206,23 @@ class ImportScripts::Base
next # block returns nil to skip a post
end
if user_id_from_imported_user_id(u[:id])
import_id = u[:id]
if user_id_from_imported_user_id(import_id)
users_skipped += 1
elsif u[:email].present?
new_user = create_user(u, u[:id])
new_user = create_user(u, import_id)
if new_user.valid?
@existing_users[u[:id].to_s] = new_user.id
@existing_users[import_id.to_s] = new_user.id
users_created += 1
else
@failed_users << u
puts "Failed to create user id: #{u[:id]}, username: #{new_user.username}, email: #{new_user.email}: #{new_user.errors.full_messages}"
puts "Failed to create user id: #{import_id}, username: #{new_user.username}, email: #{new_user.email}: #{new_user.errors.full_messages}"
end
else
@failed_users << u
puts "Skipping user id #{u[:id]} because email is blank"
puts "Skipping user id #{import_id} because email is blank"
end
print_status users_created + users_skipped + @failed_users.length + (opts[:offset] || 0), total
@ -232,6 +238,7 @@ class ImportScripts::Base
return existing if existing && existing.custom_fields["import_id"].to_i == import_id.to_i
bio_raw = opts.delete(:bio_raw)
website = opts.delete(:website)
avatar_url = opts.delete(:avatar_url)
opts[:name] = User.suggest_name(opts[:email]) unless opts[:name]
@ -256,8 +263,9 @@ class ImportScripts::Base
begin
User.transaction do
u.save!
if bio_raw.present?
u.user_profile.bio_raw = bio_raw
if bio_raw.present? || website.present?
u.user_profile.bio_raw = bio_raw if bio_raw.present?
u.user_profile.website = website if website.present?
u.user_profile.save!
end
end
@ -284,7 +292,7 @@ class ImportScripts::Base
def create_categories(results)
results.each do |c|
params = yield(c)
puts " #{params[:name]}"
puts "\t#{params[:name]}"
# make sure categories don't go more than 2 levels deep
if params[:parent_category_id]
@ -299,10 +307,11 @@ class ImportScripts::Base
end
def create_category(opts, import_id)
existing = category_from_imported_category_id(import_id)
existing = category_from_imported_category_id(import_id) || Category.where("LOWER(name) = ?", opts[:name].downcase).first
return existing if existing
post_create_action = opts.delete(:post_create_action)
new_category = Category.new(
name: opts[:name],
user_id: opts[:user_id] || opts[:user].try(:id) || -1,
@ -310,9 +319,12 @@ class ImportScripts::Base
description: opts[:description],
parent_category_id: opts[:parent_category_id]
)
new_category.custom_fields["import_id"] = import_id if import_id
new_category.save!
post_create_action.try(:call, new_category)
new_category
end
@ -343,7 +355,11 @@ class ImportScripts::Base
new_post = create_post(params, import_id)
if new_post.is_a?(Post)
@existing_posts[import_id] = new_post.id
@topic_lookup[new_post.id] = {post_number: new_post.post_number, topic_id: new_post.topic_id}
@topic_lookup[new_post.id] = {
post_number: new_post.post_number,
topic_id: new_post.topic_id,
url: new_post.url,
}
created += 1
else
@ -351,13 +367,13 @@ class ImportScripts::Base
puts "Error creating post #{import_id}. Skipping."
puts new_post.inspect
end
rescue Discourse::InvalidAccess => e
skipped += 1
puts "InvalidAccess creating post #{import_id}. Topic is closed? #{e.message}"
rescue => e
skipped += 1
puts "Exception while creating post #{import_id}. Skipping."
puts e.message
rescue Discourse::InvalidAccess => e
skipped += 1
puts "InvalidAccess creating post #{import_id}. Topic is closed? #{e.message}"
end
end

592
script/import_scripts/vbulletin.rb
View File

@ -1 +1,591 @@
# TODO: migrate it from https://github.com/discourse/discourse/blob/import-vbulletin/script/import_scripts/vbulletin.rb
require File.expand_path(File.dirname(__FILE__) + "/base.rb")
require "optparse"
require "csv"
class ImportScripts::VBulletin < ImportScripts::Base
attr_reader :options
def self.run
options = Options.new
begin
options.parse!
rescue OptionParser::MissingArgument, OptionParser::InvalidArgument => e
$stderr.puts e.to_s.capitalize
$stderr.puts options.usage
exit 1
end
new(options).perform
end
def initialize(options)
super()
@options = options
end
def execute
load_groups_mapping
load_groups
load_users
load_categories_mappings
load_categories
load_categories_permissions
load_topics
load_posts
import_groups
import_users
create_groups_membership
import_categories
# import_category_groups
preprocess_posts
import_topics
import_posts
postprocess_posts
puts "", "Done"
end
private
############################################################################
# LOAD #
############################################################################
def default_csv_options
@@default_csv_options ||= { headers: true, header_converters: :symbol }
end
def load_groups_mapping
return if @options.group_mapping.blank?
puts "", "Loading groups mappings from '#{@options.group_mapping}'..."
data = File.read(@options.group_mapping)
@groups_mappings = CSV.parse(data, default_csv_options).map { |row| row.to_hash }
@mapped_groups = {}
@new_group_ids = {}
@groups_mappings.each do |gm|
@mapped_groups[gm[:old_id]] = gm
@new_group_ids[gm[:new_id]] ||= true
end
puts "Loaded #{@groups_mappings.count} groups mappings for #{@new_group_ids.count} groups!"
end
def load_groups
puts "", "Loading groups from '#{@options.user_group}'..."
data = File.read(@options.user_group)
@groups = CSV.parse(data, default_csv_options).map { |row| row.to_hash }
original_count = @groups.count
# reject unmapped groups
if @new_group_ids.try(:size) > 0
@groups.reject! { |group| !@new_group_ids.has_key?(group[:usergroupid]) }
end
puts "Loaded #{@groups.count} out of #{original_count} groups!"
end
def load_users
puts "", "Loading users from '#{@options.user}'..."
data = File.read(@options.user)
csv_options = default_csv_options.merge({ col_sep: "\t", quote_char: "\u200B" })
@users = CSV.parse(data, csv_options).map { |row| row.to_hash }
original_count = @users.count
if @mapped_groups.try(:size) > 0
# reject users from unmapped groups
@users.reject! { |user| !@mapped_groups.has_key?(user[:usergroupid]) }
# change mapped groups
@users.each { |user| user[:usergroupid] = @mapped_groups[user[:usergroupid]][:new_id] }
end
puts "Loaded #{@users.count} out of #{original_count} users!"
end
def load_categories_mappings
return if @options.forum_mapping.blank?
puts "", "Loading categories mappings from '#{@options.forum_mapping}'..."
data = File.read(@options.forum_mapping)
@categories_mappings = CSV.parse(data, default_csv_options).map { |row| row.to_hash }
@mapped_categories = {}
@new_category_ids = {}
@categories_mappings.each do |cm|
@mapped_categories[cm[:old_id]] = cm
@new_category_ids[cm[:new_id]] ||= true
end
puts "Loaded #{@categories_mappings.count} categories mappings for #{@new_category_ids.count} categories!"
end
def load_categories
puts "", "Loading categories from '#{@options.forum}'..."
data = File.read(@options.forum)
@categories = CSV.parse(data, default_csv_options).map { |row| row.to_hash }
original_count = @categories.count
if @new_category_ids.try(:size) > 0
# reject unmapped categories
@categories.reject! { |category| !@new_category_ids.has_key?(category[:forumid]) }
# update mapped categories' title
@categories.each { |category| category[:title] = @mapped_categories[category[:forumid]][:new_name] }
end
puts "Loaded #{@categories.count} out of #{original_count} categories!"
end
# extracted from the "bitfield_vbulletin.xml" file
VB_FORUM_PERMISSIONS_CAN_VIEW = 1
VB_FORUM_PERMISSIONS_CAN_VIEW_THREADS = 524288
VB_FORUM_PERMISSIONS_CAN_REPLY_OWN = 32
VB_FORUM_PERMISSIONS_CAN_REPLY_OTHERS = 64
VB_FORUM_PERMISSIONS_CAN_POST_NEW = 16
def translate_forum_permissions(permissions)
can_see = ((permissions & VB_FORUM_PERMISSIONS_CAN_VIEW) | (permissions & VB_FORUM_PERMISSIONS_CAN_VIEW_THREADS)) > 0
can_reply = ((permissions & VB_FORUM_PERMISSIONS_CAN_REPLY_OWN) | (permissions & VB_FORUM_PERMISSIONS_CAN_REPLY_OTHERS)) > 0
can_create = (permissions & VB_FORUM_PERMISSIONS_CAN_POST_NEW) > 0
return CategoryGroup.permission_types[:full] if can_create
return CategoryGroup.permission_types[:create_post] if can_reply
return CategoryGroup.permission_types[:readonly] if can_see
nil
end
def load_categories_permissions
puts "", "Loading categories permissions from '#{@options.forum_permission}'..."
data = File.read(@options.forum_permission)
@categories_permissions = CSV.parse(data, default_csv_options).map { |row| row.to_hash }
original_count = @categories_permissions.count
# reject unmapped groups
if @mapped_groups.try(:size) > 0
@categories_permissions.reject! { |cp| !@mapped_groups.has_key?(cp[:usergroupid]) }
end
# reject unmapped categories
if @mapped_categories.try(:size) > 0
@categories_permissions.reject! { |cp| !@mapped_categories.has_key?(cp[:forumid]) }
end
# translate permissions
@categories_permissions.each do |cp|
cp[:permission] = translate_forum_permissions(cp[:forumpermissions].to_i)
cp[:usergroupid] = @mapped_groups[cp[:usergroupid]][:new_id]
cp[:forumid] = @mapped_categories[cp[:forumid]][:new_id]
end
# clean permissions up
@categories_permissions.reject! { |cp| cp[:permission].nil? }
puts "Loaded #{@categories_permissions.count} out of #{original_count} categories permissions!"
end
def load_topics
puts "", "Loading topics from '#{@options.thread}'..."
data = File.read(@options.thread)
csv_options = default_csv_options.merge({ col_sep: "\t", quote_char: "\u200B" })
@topics = CSV.parse(data, csv_options).map { |row| row.to_hash }
original_count = @topics.count
if @mapped_categories.try(:size) > 0
# reject topics from unmapped categories
@topics.reject! { |topic| !@mapped_categories.has_key?(topic[:forumid]) }
# change mapped categories
@topics.each do |topic|
topic[:old_forumid] = topic[:forumid]
topic[:forumid] = @mapped_categories[topic[:forumid]][:new_id]
end
end
puts "Loaded #{@topics.count} out of #{original_count} topics!"
end
def load_posts
puts "", "Loading posts from '#{@options.post}'..."
data = File.read(@options.post)
csv_options = default_csv_options.merge({ col_sep: "\t", quote_char: "\u200B" })
@posts = CSV.parse(data, csv_options).map { |row| row.to_hash }
original_count = @posts.count
# reject posts without topics
topic_ids = Set.new(@topics.map { |t| t[:threadid] })
@posts.reject! { |post| !topic_ids.include?(post[:threadid]) }
puts "Loaded #{@posts.count} out of #{original_count} posts!"
end
############################################################################
# IMPORT #
############################################################################
def import_groups
puts "", "Importing groups..."
create_groups(@groups) do |group|
{
id: group[:usergroupid],
name: group[:title],
}
end
end
def import_users
puts "", "Importing users..."
@old_username_to_new_usernames = {}
create_users(@users) do |user|
@old_username_to_new_usernames[user[:username]] = UserNameSuggester.suggest(user[:username])
{
id: user[:userid],
username: @old_username_to_new_usernames[user[:username]],
email: user[:email],
website: user[:homepage],
title: user[:usertitle],
primary_group_id: group_id_from_imported_group_id(user[:usergroupid]),
}
end
end
def create_groups_membership
puts "", "Creating groups membership..."
Group.find_each do |group|
next if group.automatic
puts "\t#{group.name}"
next if GroupUser.where(group_id: group.id).count > 0
user_ids_in_group = User.where(primary_group_id: group.id).pluck(:id).to_a
next if user_ids_in_group.size == 0
values = user_ids_in_group.map { |user_id| "(#{group.id}, #{user_id})" }.join(",")
User.exec_sql <<-SQL
BEGIN;
INSERT INTO group_users (group_id, user_id) VALUES #{values};
COMMIT;
SQL
Group.reset_counters(group.id, :group_users)
end
end
def import_categories
puts "", "Importing categories..."
create_categories(@categories) do |category|
{
id: category[:forumid],
name: category[:title].strip[0...50],
position: category[:displayorder].to_i,
description: category[:description],
}
end
end
def import_category_groups
puts "", "Importing category groups..."
# TODO
end
def preprocess_posts
puts "", "Preprocessing posts..."
current = 0
max = @posts.size
@posts.each do |post|
post[:raw] = preprocess_post_raw(post[:pagetext])
current += 1
print_status(current, max)
end
end
def preprocess_post_raw(raw)
return "" if raw.blank?
raw = raw.gsub(/(\\r)?\\n/, "\n")
.gsub("\\t", "\t")
# [URL]...[/URL]
raw = raw.gsub(/\[url\](.+?)\[\/url\]/i) { $1.to_s }
# [MENTION]<username>[/MENTION]
raw = raw.gsub(/\[mention\](.+?)\[\/mention\]/i) do
old_username = $1
if @old_username_to_new_usernames.has_key?(old_username)
username = @old_username_to_new_usernames[old_username]
"@#{username}"
else
$&
end
end
# [MENTION=<user_id>]...[/MENTION]
raw = raw.gsub(/\[mention=(\d+)\].+?\[\/mention\]/i) do
user_id = $1
if user = @users.select { |u| u[:userid] == user_id }.first
username = @old_username_to_new_usernames[user[:username]] || user[:username]
"@#{username}"
else
$&
end
end
# [QUOTE]...[/QUOTE]
raw = raw.gsub(/\[quote\](.+?)\[\/quote\]/im) { "\n> #{$1}\n" }
# [QUOTE=<username>]...[/QUOTE]
raw = raw.gsub(/\[quote=([^;\]]+)\](.+?)\[\/quote\]/im) do
old_username, quote = $1, $2
if @old_username_to_new_usernames.has_key?(old_username)
username = @old_username_to_new_usernames[old_username]
"\n[quote=\"#{username}\"]\n#{quote}\n[/quote]\n"
else
$&
end
end
# [HTML]...[/HTML]
# [PHP]...[/PHP]
["html", "php"].each do |language|
raw = raw.gsub(/\[#{language}\](.+?)\[\/#{language}\]/im) { "\n```#{language}\n#{$1}\n```\n" }
end
# [CODE]...[/CODE]
raw = raw.gsub(/\[code\](.+?)\[\/code\]/im) { "\n```\n#{$1}\n```\n" }
# [HIGHLIGHT="..."]...[/HIGHLIGHT]
raw = raw.gsub(/\[highlight(?:[^\]]*)\](.+)\[\/highlight\]/im) { "\n```\n#{$1}\n```\n" }
# [SAMP]...[SAMP]
raw = raw.gsub(/\[samp\](.+?)\[\/samp\]/i) { "`#{$1}`" }
# [YOUTUBE]<id>[/YOUTUBE]
raw = raw.gsub(/\[youtube\](.+?)\[\/youtube\]/i) { "\n//youtu.be/#{$1}\n" }
# [VIDEO=youtube;<id>]...[/VIDEO]
raw = raw.gsub(/\[video=youtube;([^\]]+)\].*\[\/video\]/i) { "\n//youtu.be/#{$1}\n" }
# [MP3]<url>[/MP3]
raw = raw.gsub(/\[MP3\](.+?)\[\/MP3\]/i) { "\n#{$1}\n" }
raw
end
def import_topics
puts "", "Importing topics..."
create_posts(@topics) do |topic|
next unless post = @posts.select { |p| p[:postid] == topic[:firstpostid] }.first
t = {
id: "thread#" + topic[:threadid],
user_id: user_id_from_imported_user_id(topic[:postuserid]) || Discourse::SYSTEM_USER_ID,
title: CGI.unescapeHTML(topic[:title]).strip[0...255],
category: category_from_imported_category_id(topic[:forumid]).try(:name),
raw: post[:raw],
created_at: Time.at(topic[:dateline].to_i),
visible: topic[:visible].to_i == 1,
closed: topic[:open].to_i == 0,
views: topic[:views].to_i,
}
if topic[:sticky].to_i == 1
t[:pinned_at] = t[:created_at]
end
# tag
if (tag = @mapped_categories[topic[:old_forumid]][:tag] || "").present?
t[:custom_fields] ||= {}
t[:custom_fields]['tag'] = tag
end
t
end
end
def import_posts
puts "", "Importing posts..."
# reject all first posts
first_post_ids = Set.new(@topics.map { |t| t[:firstpostid] })
@posts.reject! { |post| first_post_ids.include?(post[:postid]) }
create_posts(@posts) do |post|
t = topic_lookup_from_imported_post_id("thread#" + post[:threadid])
p = {
id: post[:postid],
user_id: user_id_from_imported_user_id(post[:postuserid]) || Discourse::SYSTEM_USER_ID,
topic_id: t[:topic_id],
raw: post[:raw],
created_at: Time.at(post[:dateline].to_i),
hidden: post[:visible].to_i == 0,
}
if (edit_reason = (post[:editreason] || "").gsub("NULL", "")).present?
p[:edit_reason] = edit_reason
end
if parent = topic_lookup_from_imported_post_id(post[:parentid])
p[:reply_to_post_number] = parent[:post_number]
end
p
end
end
def postprocess_posts
puts "", "Postprocessing posts..."
current = 0
max = @posts.size
@posts.each do |post|
new_raw = postprocess_post_raw(post[:raw])
if new_raw != post[:raw]
p = Post.find(post_id_from_imported_post_id(post[:postid]))
p.raw = new_raw
p.save
end
current += 1
print_status(current, max)
end
end
def postprocess_post_raw(raw)
# [QUOTE=<username>;<post_id>]...[/QUOTE]
raw = raw.gsub(/\[quote=([^;]+);(\d+)\](.+?)\[\/quote\]/im) do
old_username, post_id, quote = $1, $2, $3
if @old_username_to_new_usernames.has_key?(old_username) && topic_lookup = topic_lookup_from_imported_post_id(post_id)
post_number = topic_lookup[:post_number]
topic_id = topic_lookup[:topic_id]
username = @old_username_to_new_usernames[old_username]
"\n[quote=\"#{username},post:#{post_number},topic:#{topic_id}\"]\n#{quote}\n[/quote]\n"
else
$&
end
end
# [THREAD]<thread_id>[/THREAD]
# ==> http://my.discourse.org/t/slug/<topic_id>
raw = raw.gsub(/\[thread\](\d+)\[\/thread\]/i) do
thread_id = $1
if topic_lookup = topic_lookup_from_imported_post_id("thread#" + thread_id)
topic_lookup[:url]
else
$&
end
end
# [THREAD=<thread_id>]...[/THREAD]
# ==> [...](http://my.discourse.org/t/slug/<topic_id>)
raw = raw.gsub(/\[thread=(\d+)\](.+?)\[\/thread\]/i) do
thread_id, link = $1, $2
if topic_lookup = topic_lookup_from_imported_post_id("thread#" + thread_id)
url = topic_lookup[:url]
"[#{link}](#{url})"
else
$&
end
end
# [POST]<post_id>[/POST]
# ==> http://my.discourse.org/t/slug/<topic_id>/<post_number>
raw = raw.gsub(/\[post\](\d+)\[\/post\]/i) do
post_id = $1
if topic_lookup = topic_lookup_from_imported_post_id(post_id)
topic_lookup[:url]
else
$&
end
end
# [POST=<post_id>]...[/POST]
# ==> [...](http://my.discourse.org/t/<topic_slug>/<topic_id>/<post_number>)
raw = raw.gsub(/\[post=(\d+)\](.+?)\[\/post\]/i) do
post_id, link = $1, $2
if topic_lookup = topic_lookup_from_imported_post_id(post_id)
url = topic_lookup[:url]
"[#{link}](#{url})"
else
$&
end
end
raw
end
############################################################################
# OPTIONS #
############################################################################
class Options
attr_accessor :user_group, :user, :forum, :forum_permission, :thread, :post
attr_accessor :group_mapping, :forum_mapping
def parse!(args = ARGV)
parser.parse!(args)
[:user_group, :user, :forum, :forum_permission, :thread, :post].each do |option_name|
option = self.send(option_name)
raise OptionParser::MissingArgument.new(option_name) if option.nil?
raise OptionParser::InvalidArgument.new("#{option} file does not exist") if !File.exists?(option)
end
end
def usage
parser.to_s
end
private
def parser
@parser ||= OptionParser.new(nil, 50) do |opts|
opts.banner = "Usage:\truby #{File.basename($0)} [options]"
opts.on("--user-group USER-GROUP.csv", "list of usergroups") { |s| self.user_group = s }
opts.on("--user USER.csv", "list of users") { |s| self.user = s }
opts.on("--forum FORUM.csv", "list of forums") { |s| self.forum = s }
opts.on("--forum-permission FORUM-PERMISSION.csv", "list of forum permissions") { |s| self.forum_permission = s }
opts.on("--thread THREAD.csv", "list of threads") { |s| self.thread = s }
opts.on("--post POST.csv", "list of posts") { |s| self.post = s }
opts.on("--group-mapping GROUP-MAPPING.csv", "list of group mappings") { |s| self.group_mapping = s }
opts.on("--forum-mapping FORUM-MAPPING.csv", "list of forum mappings") { |s| self.forum_mapping = s }
end
end
end
end
ImportScripts::VBulletin.run