discourse-data-explorer/plugin.rb

1168 lines
37 KiB
Ruby
Raw Normal View History

# name: discourse-data-explorer
# about: Interface for running analysis SQL queries on the live database
2015-06-30 18:14:24 -04:00
# version: 0.2
# authors: Riking
# url: https://github.com/discourse/discourse-data-explorer
enabled_site_setting :data_explorer_enabled
2015-06-30 13:37:48 -04:00
register_asset 'stylesheets/explorer.scss'
2018-11-08 12:01:50 -05:00
2018-11-15 11:12:32 -05:00
if respond_to?(:register_svg_icon)
register_svg_icon "caret-down"
register_svg_icon "caret-right"
register_svg_icon "chevron-left"
register_svg_icon "exclamation-circle"
register_svg_icon "info"
register_svg_icon "pencil-alt"
register_svg_icon "upload"
end
# route: /admin/plugins/explorer
add_admin_route 'explorer.title', 'explorer'
module ::DataExplorer
QUERY_RESULT_MAX_LIMIT = 1000
def self.plugin_name
'discourse-data-explorer'.freeze
end
def self.pstore_get(key)
PluginStore.get(DataExplorer.plugin_name, key)
end
def self.pstore_set(key, value)
PluginStore.set(DataExplorer.plugin_name, key, value)
end
def self.pstore_delete(key)
PluginStore.remove(DataExplorer.plugin_name, key)
end
end
after_initialize do
module ::DataExplorer
class Engine < ::Rails::Engine
engine_name "data_explorer"
isolate_namespace DataExplorer
end
class ValidationError < StandardError
2015-07-08 16:45:13 -04:00
end
2015-08-05 19:40:00 -04:00
class SmallBadgeSerializer < ApplicationSerializer
attributes :id, :name, :badge_type, :description, :icon
end
class SmallPostWithExcerptSerializer < ApplicationSerializer
attributes :id, :topic_id, :post_number, :excerpt
2015-09-14 18:34:57 -04:00
attributes :username, :avatar_template
2015-08-05 19:40:00 -04:00
def excerpt
Post.excerpt(object.cooked, 70)
end
def username
object.user && object.user.username
end
def avatar_template
object.user && object.user.avatar_template
end
2015-08-05 19:40:00 -04:00
end
# Run a data explorer query on the currently connected database.
#
# @param [DataExplorer::Query] query the Query object to run
# @param [Hash] params the colon-style query parameters to pass to AR
# @param [Hash] opts hash of options
# explain - include a query plan in the result
# @return [Hash]
# error - any exception that was raised in the execution. Check this
# first before looking at any other fields.
# pg_result - the PG::Result object
# duration_nanos - the query duration, in nanoseconds
# explain - the query
2017-08-02 01:42:49 -04:00
def self.run_query(query, req_params = {}, opts = {})
# Safety checks
2015-07-15 16:20:42 -04:00
# see test 'doesn't allow you to modify the database #2'
if query.sql =~ /;/
err = DataExplorer::ValidationError.new(I18n.t('js.errors.explorer.no_semicolons'))
2017-08-02 01:42:49 -04:00
return { error: err, duration_nanos: 0 }
end
query_args = {}
begin
query_args = query.cast_params req_params
rescue DataExplorer::ValidationError => e
2017-08-02 01:42:49 -04:00
return { error: e, duration_nanos: 0 }
2015-06-30 22:51:38 -04:00
end
time_start, time_end, explain, err, result = nil
begin
ActiveRecord::Base.connection.transaction do
# Setting transaction to read only prevents shoot-in-foot actions like SELECT FOR UPDATE
2015-07-15 16:20:42 -04:00
# see test 'doesn't allow you to modify the database #1'
DB.exec "SET TRANSACTION READ ONLY"
# Set a statement timeout so we can't tie up the server
DB.exec "SET LOCAL statement_timeout = 10000"
2015-06-25 17:55:06 -04:00
2015-07-15 16:20:42 -04:00
# SQL comments are for the benefits of the slow queries log
sql = <<-SQL
/*
2015-06-25 17:55:06 -04:00
* DataExplorer Query
2015-06-30 22:51:38 -04:00
* Query: /admin/plugins/explorer?id=#{query.id}
2015-06-25 17:55:06 -04:00
* Started by: #{opts[:current_user]}
*/
WITH query AS (
#{query.sql}
) SELECT * FROM query
LIMIT #{opts[:limit] || DataExplorer::QUERY_RESULT_MAX_LIMIT}
SQL
time_start = Time.now
# we probably want to rewrite this ... but for now reuse the working
# code we have
sql = DB.param_encoder.encode(sql, query_args)
result = ActiveRecord::Base.connection.raw_connection.async_exec(sql)
result.check # make sure it's done
time_end = Time.now
if opts[:explain]
explain = DB.query_hash("EXPLAIN #{query.sql}", query_args)
2017-08-02 01:42:49 -04:00
.map { |row| row["QUERY PLAN"] }.join "\n"
end
# All done. Issue a rollback anyways, just in case
2015-07-15 16:20:42 -04:00
# see test 'doesn't allow you to modify the database #1'
raise ActiveRecord::Rollback
end
rescue Exception => ex
err = ex
time_end = Time.now
end
{
error: err,
pg_result: result,
duration_secs: time_end - time_start,
explain: explain,
params_full: query_args
}
end
2015-08-05 19:40:00 -04:00
def self.extra_data_pluck_fields
@extra_data_pluck_fields ||= {
2017-08-02 01:42:49 -04:00
user: { class: User, fields: [:id, :username, :uploaded_avatar_id], serializer: BasicUserSerializer },
badge: { class: Badge, fields: [:id, :name, :badge_type_id, :description, :icon], include: [:badge_type], serializer: SmallBadgeSerializer },
post: { class: Post, fields: [:id, :topic_id, :post_number, :cooked, :user_id], include: [:user], serializer: SmallPostWithExcerptSerializer },
topic: { class: Topic, fields: [:id, :title, :slug, :posts_count], serializer: BasicTopicSerializer },
group: { class: Group, ignore: true },
category: { class: Category, ignore: true },
reltime: { ignore: true },
html: { ignore: true },
2015-08-05 19:40:00 -04:00
}
end
2015-08-25 23:48:19 -04:00
def self.column_regexes
@column_regexes ||=
extra_data_pluck_fields.map do |key, val|
if val[:class]
/(#{val[:class].to_s.downcase})_id$/
end
end.compact
end
2015-08-05 19:40:00 -04:00
def self.add_extra_data(pg_result)
needed_classes = {}
pg_result.fields.each_with_index do |col, idx|
2018-10-22 16:24:55 -04:00
rgx = column_regexes.find { |r| r.match col }
2015-08-25 23:48:19 -04:00
if rgx
cls = (rgx.match col)[1].to_sym
needed_classes[cls] ||= []
needed_classes[cls] << idx
elsif col =~ /^(\w+)\$/
cls = $1.to_sym
needed_classes[cls] ||= []
needed_classes[cls] << idx
2015-08-05 19:40:00 -04:00
end
end
ret = {}
2015-08-25 23:48:19 -04:00
col_map = {}
2015-08-05 19:40:00 -04:00
needed_classes.each do |cls, column_nums|
next unless column_nums.present?
2015-08-25 23:48:19 -04:00
support_info = extra_data_pluck_fields[cls]
next unless support_info
column_nums.each do |col_n|
col_map[col_n] = cls
end
if support_info[:ignore]
ret[cls] = []
next
end
2015-08-05 19:40:00 -04:00
ids = Set.new
column_nums.each do |col_n|
ids.merge(pg_result.column_values(col_n))
end
ids.delete nil
ids.map! &:to_i
object_class = support_info[:class]
all_objs = object_class
all_objs = all_objs.with_deleted if all_objs.respond_to? :with_deleted
all_objs = all_objs
2017-08-02 01:42:49 -04:00
.select(support_info[:fields])
.where(id: ids.to_a.sort)
.includes(support_info[:include])
.order(:id)
2015-08-05 19:40:00 -04:00
ret[cls] = ActiveModel::ArraySerializer.new(all_objs, each_serializer: support_info[:serializer])
end
2015-08-25 23:48:19 -04:00
[ret, col_map]
2015-08-05 19:40:00 -04:00
end
2015-07-08 16:45:13 -04:00
def self.sensitive_column_names
%w(
#_IP_Addresses
topic_views.ip_address
users.ip_address
users.registration_ip_address
incoming_links.ip_address
topic_link_clicks.ip_address
user_histories.ip_address
#_Emails
email_tokens.email
users.email
invites.email
user_histories.email
email_logs.to_address
posts.raw_email
badge_posts.raw_email
#_Secret_Tokens
email_tokens.token
email_logs.reply_key
api_keys.key
site_settings.value
users.auth_token
users.password_hash
users.salt
#_Authentication_Info
user_open_ids.email
oauth2_user_infos.uid
oauth2_user_infos.email
facebook_user_infos.facebook_user_id
facebook_user_infos.email
twitter_user_infos.twitter_user_id
github_user_infos.github_user_id
single_sign_on_records.external_email
single_sign_on_records.external_id
google_user_infos.google_user_id
google_user_infos.email
2015-07-08 16:45:13 -04:00
)
end
def self.schema
2015-07-28 14:18:22 -04:00
# No need to expire this, because the server processes get restarted on upgrade
2015-07-08 16:45:13 -04:00
# refer user to http://www.postgresql.org/docs/9.3/static/datatype.html
@schema ||= begin
results = DB.query_hash <<~SQL
select
c.column_name column_name,
c.data_type data_type,
c.character_maximum_length character_maximum_length,
c.is_nullable is_nullable,
c.column_default column_default,
c.table_name table_name,
pgd.description column_desc
from INFORMATION_SCHEMA.COLUMNS c
inner join pg_catalog.pg_statio_all_tables st on (c.table_schema = st.schemaname and c.table_name = st.relname)
left outer join pg_catalog.pg_description pgd on (pgd.objoid = st.relid and pgd.objsubid = c.ordinal_position)
where c.table_schema = 'public'
ORDER BY c.table_name, c.ordinal_position
SQL
2015-07-08 16:45:13 -04:00
by_table = {}
# Massage the results into a nicer form
results.each do |hash|
2015-07-09 18:46:19 -04:00
full_col_name = "#{hash['table_name']}.#{hash['column_name']}"
2015-07-08 16:45:13 -04:00
if hash['is_nullable'] == "YES"
hash['is_nullable'] = true
else
hash.delete('is_nullable')
end
clen = hash.delete 'character_maximum_length'
dt = hash['data_type']
2015-07-28 14:18:22 -04:00
if hash['column_name'] == 'id'
hash['data_type'] = 'serial'
hash['primary'] = true
elsif dt == 'character varying'
2015-07-08 16:45:13 -04:00
hash['data_type'] = "varchar(#{clen.to_i})"
elsif dt == 'timestamp without time zone'
hash['data_type'] = 'timestamp'
elsif dt == 'double precision'
hash['data_type'] = 'double'
end
default = hash['column_default']
if default.nil? || default =~ /^nextval\(/
hash.delete 'column_default'
elsif default =~ /^'(.*)'::(character varying|text)/
hash['column_default'] = $1
end
2015-07-27 19:05:10 -04:00
hash.delete('column_desc') unless hash['column_desc']
2015-07-08 16:45:13 -04:00
2015-07-09 18:46:19 -04:00
if sensitive_column_names.include? full_col_name
2015-07-08 16:45:13 -04:00
hash['sensitive'] = true
end
2015-07-09 18:46:19 -04:00
if enum_info.include? full_col_name
hash['enum'] = enum_info[full_col_name]
end
2015-07-28 14:18:22 -04:00
if denormalized_columns.include? full_col_name
hash['denormal'] = denormalized_columns[full_col_name]
end
fkey = fkey_info(hash['table_name'], hash['column_name'])
if fkey
hash['fkey_info'] = fkey
end
2015-07-08 16:45:13 -04:00
table_name = hash.delete('table_name')
by_table[table_name] ||= []
by_table[table_name] << hash
2015-07-08 16:45:13 -04:00
end
# this works for now, but no big loss if the tables aren't quite sorted
2015-07-09 17:22:54 -04:00
favored_order = %w(posts topics users categories badges groups notifications post_actions site_settings)
2015-07-08 16:45:13 -04:00
sorted_by_table = {}
2015-07-09 17:22:54 -04:00
favored_order.each do |tbl|
sorted_by_table[tbl] = by_table[tbl]
end
2015-07-08 16:45:13 -04:00
by_table.keys.sort.each do |tbl|
2015-07-09 17:22:54 -04:00
next if favored_order.include? tbl
2015-07-08 16:45:13 -04:00
sorted_by_table[tbl] = by_table[tbl]
end
sorted_by_table
end
end
2015-07-09 18:46:19 -04:00
def self.enums
@enums ||= {
2017-08-02 01:42:49 -04:00
'badges.badge_type_id': Enum.new(:gold, :silver, :bronze, start: 1),
'category_groups.permission_type': CategoryGroup.permission_types,
'directory_items.period_type': DirectoryItem.period_types,
'groups.alias_level': Group::ALIAS_LEVELS,
'groups.id': Group::AUTO_GROUPS,
'notifications.notification_type': Notification.types,
'posts.cook_method': Post.cook_methods,
'posts.hidden_reason_id': Post.hidden_reasons,
'posts.post_type': Post.types,
'post_actions.post_action_type_id': PostActionType.types,
'post_action_types.id': PostActionType.types,
'queued_posts.state': QueuedPost.states,
'site_settings.data_type': SiteSetting.types,
'topic_users.notification_level': TopicUser.notification_levels,
'topic_users.notifications_reason_id': TopicUser.notification_reasons,
'user_histories.action': UserHistory.actions,
'users.trust_level': TrustLevel.levels,
2015-07-09 18:46:19 -04:00
}.with_indifferent_access
end
def self.enum_info
@enum_info ||= begin
enum_info = {}
2017-08-02 01:42:49 -04:00
enums.map do |key, enum|
2015-07-09 18:46:19 -04:00
# https://stackoverflow.com/questions/10874356/reverse-a-hash-in-ruby
enum_info[key] = Hash[enum.to_a.map(&:reverse)]
end
enum_info
end
end
def self.fkey_info(table, column)
full_name = "#{table}.#{column}"
if fkey_defaults[column]
fkey_defaults[column]
elsif column =~ /_by_id$/ || column =~ /_user_id$/
:users
elsif foreign_keys[full_name]
foreign_keys[full_name]
else
nil
end
end
def self.foreign_keys
@fkey_columns ||= {
2017-08-02 01:42:49 -04:00
'posts.last_editor_id': :users,
'posts.version': :'post_revisions.number',
2017-08-02 01:42:49 -04:00
'topics.featured_user1_id': :users,
'topics.featured_user2_id': :users,
'topics.featured_user3_id': :users,
'topics.featured_user4_id': :users,
'topics.featured_user5_id': :users,
2017-08-02 01:42:49 -04:00
'users.seen_notification_id': :notifications,
'users.uploaded_avatar_id': :uploads,
'users.primary_group_id': :groups,
2017-08-02 01:42:49 -04:00
'categories.latest_post_id': :posts,
'categories.latest_topic_id': :topics,
'categories.parent_category_id': :categories,
2017-08-02 01:42:49 -04:00
'badges.badge_grouping_id': :badge_groupings,
2017-08-02 01:42:49 -04:00
'post_actions.related_post_id': :posts,
2017-08-02 01:42:49 -04:00
'color_scheme_colors.color_scheme_id': :color_schemes,
'color_schemes.versioned_id': :color_schemes,
2017-08-02 01:42:49 -04:00
'incoming_links.incoming_referer_id': :incoming_referers,
'incoming_referers.incoming_domain_id': :incoming_domains,
2017-08-02 01:42:49 -04:00
'post_replies.reply_id': :posts,
2017-08-02 01:42:49 -04:00
'quoted_posts.quoted_post_id': :posts,
2017-08-02 01:42:49 -04:00
'topic_link_clicks.topic_link_id': :topic_links,
'topic_link_clicks.link_topic_id': :topics,
'topic_link_clicks.link_post_id': :posts,
2017-08-02 01:42:49 -04:00
'user_actions.target_topic_id': :topics,
'user_actions.target_post_id': :posts,
2017-08-02 01:42:49 -04:00
'user_avatars.custom_upload_id': :uploads,
'user_avatars.gravatar_upload_id': :uploads,
2017-08-02 01:42:49 -04:00
'user_badges.notification_id': :notifications,
2017-08-02 01:42:49 -04:00
'user_profiles.card_image_badge_id': :badges,
}.with_indifferent_access
end
def self.fkey_defaults
@fkey_defaults ||= {
2017-08-02 01:42:49 -04:00
user_id: :users,
# :*_by_id => :users,
# :*_user_id => :users,
2017-08-02 01:42:49 -04:00
category_id: :categories,
group_id: :groups,
post_id: :posts,
post_action_id: :post_actions,
topic_id: :topics,
upload_id: :uploads,
}.with_indifferent_access
end
2015-07-28 14:18:22 -04:00
def self.denormalized_columns
{
2017-08-02 01:42:49 -04:00
'posts.reply_count': :post_replies,
'posts.quote_count': :quoted_posts,
'posts.incoming_link_count': :topic_links,
'posts.word_count': :posts,
'posts.avg_time': :post_timings,
'posts.reads': :post_timings,
'posts.like_score': :post_actions,
'posts.like_count': :post_actions,
'posts.bookmark_count': :post_actions,
'posts.vote_count': :post_actions,
'posts.off_topic_count': :post_actions,
'posts.notify_moderators_count': :post_actions,
'posts.spam_count': :post_actions,
'posts.illegal_count': :post_actions,
'posts.inappropriate_count': :post_actions,
'posts.notify_user_count': :post_actions,
'topics.views': :topic_views,
'topics.posts_count': :posts,
'topics.reply_count': :posts,
'topics.incoming_link_count': :topic_links,
'topics.moderator_posts_count': :posts,
'topics.participant_count': :posts,
'topics.word_count': :posts,
'topics.last_posted_at': :posts,
'topics.last_post_user_idt': :posts,
'topics.avg_time': :post_timings,
'topics.highest_post_number': :posts,
'topics.image_url': :posts,
'topics.excerpt': :posts,
'topics.like_count': :post_actions,
'topics.bookmark_count': :post_actions,
'topics.vote_count': :post_actions,
'topics.off_topic_count': :post_actions,
'topics.notify_moderators_count': :post_actions,
'topics.spam_count': :post_actions,
'topics.illegal_count': :post_actions,
'topics.inappropriate_count': :post_actions,
'topics.notify_user_count': :post_actions,
'categories.topic_count': :topics,
'categories.post_count': :posts,
'categories.latest_post_id': :posts,
'categories.latest_topic_id': :topics,
'categories.description': :posts,
'categories.read_restricted': :category_groups,
'categories.topics_year': :topics,
'categories.topics_month': :topics,
'categories.topics_week': :topics,
'categories.topics_day': :topics,
'categories.posts_year': :posts,
'categories.posts_month': :posts,
'categories.posts_week': :posts,
'categories.posts_day': :posts,
'badges.grant_count': :user_badges,
'groups.user_count': :group_users,
'directory_items.likes_received': :post_actions,
'directory_items.likes_given': :post_actions,
'directory_items.topics_entered': :user_stats,
'directory_items.days_visited': :user_stats,
'directory_items.posts_read': :user_stats,
'directory_items.topic_count': :topics,
'directory_items.post_count': :posts,
'post_search_data.search_data': :posts,
'top_topics.yearly_posts_count': :posts,
'top_topics.monthly_posts_count': :posts,
'top_topics.weekly_posts_count': :posts,
'top_topics.daily_posts_count': :posts,
'top_topics.yearly_views_count': :topic_views,
'top_topics.monthly_views_count': :topic_views,
'top_topics.weekly_views_count': :topic_views,
'top_topics.daily_views_count': :topic_views,
'top_topics.yearly_likes_count': :post_actions,
'top_topics.monthly_likes_count': :post_actions,
'top_topics.weekly_likes_count': :post_actions,
'top_topics.daily_likes_count': :post_actions,
'top_topics.yearly_op_likes_count': :post_actions,
'top_topics.monthly_op_likes_count': :post_actions,
'top_topics.weekly_op_likes_count': :post_actions,
'top_topics.daily_op_likes_count': :post_actions,
'top_topics.all_score': :posts,
'top_topics.yearly_score': :posts,
'top_topics.monthly_score': :posts,
'top_topics.weekly_score': :posts,
'top_topics.daily_score': :posts,
'topic_links.clicks': :topic_link_clicks,
'topic_search_data.search_data': :topics,
'topic_users.liked': :post_actions,
'topic_users.bookmarked': :post_actions,
'user_stats.posts_read_count': :post_timings,
'user_stats.topic_reply_count': :posts,
'user_stats.first_post_created_at': :posts,
'user_stats.post_count': :posts,
'user_stats.topic_count': :topics,
'user_stats.likes_given': :post_actions,
'user_stats.likes_received': :post_actions,
'user_search_data.search_data': :user_profiles,
'users.last_posted_at': :posts,
'users.previous_visit_at': :user_visits,
2015-07-28 14:18:22 -04:00
}.with_indifferent_access
end
end
# Reimplement a couple ActiveRecord methods, but use PluginStore for storage instead
require_dependency File.expand_path('../lib/queries.rb', __FILE__)
class DataExplorer::Query
attr_accessor :id, :name, :description, :sql, :created_by, :created_at, :last_run_at
2015-06-30 15:52:17 -04:00
def initialize
@name = 'Unnamed Query'
@description = ''
2015-06-30 15:52:17 -04:00
@sql = 'SELECT 1'
end
def slug
Slug.for(name).presence || "query-#{id}"
end
def params
@params ||= DataExplorer::Parameter.create_from_sql(sql)
end
def check_params!
2015-07-14 19:01:38 -04:00
DataExplorer::Parameter.create_from_sql(sql, strict: true)
nil
end
def cast_params(input_params)
result = {}.with_indifferent_access
self.params.each do |pobj|
result[pobj.identifier] = pobj.cast_to_ruby input_params[pobj.identifier]
end
result
end
# saving/loading functions
# May want to extract this into a library or something for plugins to use?
def self.alloc_id
DistributedMutex.synchronize('data-explorer_query-id') do
max_id = DataExplorer.pstore_get("q:_id")
max_id = 1 unless max_id
DataExplorer.pstore_set("q:_id", max_id + 1)
max_id
end
end
def self.from_hash(h)
query = DataExplorer::Query.new
[:name, :description, :sql, :created_by, :created_at, :last_run_at].each do |sym|
query.send("#{sym}=", h[sym].strip) if h[sym]
end
query.id = h[:id].to_i if h[:id]
query
end
def to_hash
{
id: @id,
2015-06-30 15:52:17 -04:00
name: @name,
description: @description,
sql: @sql,
created_by: @created_by,
created_at: @created_at,
last_run_at: @last_run_at
}
end
2017-08-02 01:42:49 -04:00
def self.find(id, opts = {})
if DataExplorer.pstore_get("q:#{id}").nil? && id < 0
hash = Queries.default[id.to_s]
hash[:id] = id
from_hash hash
else
unless hash = DataExplorer.pstore_get("q:#{id}")
return DataExplorer::Query.new if opts[:ignore_deleted]
raise Discourse::NotFound
end
from_hash hash
2015-06-30 15:52:17 -04:00
end
end
def save
check_params!
@id = self.class.alloc_id unless @id && @id > 0
DataExplorer.pstore_set "q:#{id}", to_hash
end
def save_default_query
check_params!
# Read from queries.rb again to pick up any changes and save them
query = Queries.default[id.to_s]
@id = query["id"]
@sql = query["sql"]
@name = query["name"]
@description = query["description"]
DataExplorer.pstore_set "q:#{id}", to_hash
end
def destroy
DataExplorer.pstore_delete "q:#{id}"
end
def read_attribute_for_serialization(attr)
self.send(attr)
end
def self.all
PluginStoreRow.where(plugin_name: DataExplorer.plugin_name)
2015-07-08 16:45:13 -04:00
.where("key LIKE 'q:%'")
.where("key != 'q:_id'")
.map do |psr|
2018-02-23 09:01:50 -05:00
DataExplorer::Query.from_hash PluginStore.cast_value(psr.type_name, psr.value)
end.sort_by { |query| query.name }
end
2015-07-15 15:23:56 -04:00
def self.destroy_all
PluginStoreRow.where(plugin_name: DataExplorer.plugin_name)
.where("key LIKE 'q:%'")
.destroy_all
end
end
class DataExplorer::Parameter
attr_accessor :identifier, :type, :default, :nullable
def initialize(identifier, type, default, nullable)
raise DataExplorer::ValidationError.new('Parameter declaration error - identifier is missing') unless identifier
raise DataExplorer::ValidationError.new('Parameter declaration error - type is missing') unless type
# process aliases
type = type.to_sym
if DataExplorer::Parameter.type_aliases[type]
type = DataExplorer::Parameter.type_aliases[type]
end
raise DataExplorer::ValidationError.new("Parameter declaration error - unknown type #{type}") unless DataExplorer::Parameter.types[type]
@identifier = identifier
@type = type
@default = default
@nullable = nullable
begin
cast_to_ruby default unless default.blank?
rescue DataExplorer::ValidationError
raise DataExplorer::ValidationError.new("Parameter declaration error - the default value is not a valid #{type}")
end
end
def to_hash
{
identifier: @identifier,
type: @type,
default: @default,
nullable: @nullable,
}
end
def self.types
@types ||= Enum.new(
# Normal types
:int, :bigint, :boolean, :string, :date, :time, :datetime, :double,
# Selection help
:user_id, :post_id, :topic_id, :category_id, :group_id, :badge_id,
# Arrays
:int_list, :string_list, :user_list
)
end
def self.type_aliases
@type_aliases ||= {
integer: :int,
text: :string,
2015-07-14 19:01:38 -04:00
timestamp: :datetime,
}
end
def cast_to_ruby(string)
string = @default unless string
if string.blank?
if @nullable
return nil
else
raise DataExplorer::ValidationError.new("Missing parameter #{identifier} of type #{type}")
end
end
if string.downcase == '#null'
return nil
end
2017-08-02 01:42:49 -04:00
def invalid_format(string, msg = nil)
if msg
raise DataExplorer::ValidationError.new("'#{string}' is an invalid #{type} - #{msg}")
else
raise DataExplorer::ValidationError.new("'#{string}' is an invalid value for #{type}")
end
end
value = nil
case @type
2017-08-02 01:42:49 -04:00
when :int
invalid_format string, 'Not an integer' unless string =~ /^-?\d+$/
value = string.to_i
2017-09-04 02:06:19 -04:00
invalid_format string, 'Too large' unless Integer === value
2017-08-02 01:42:49 -04:00
when :bigint
invalid_format string, 'Not an integer' unless string =~ /^-?\d+$/
value = string.to_i
when :boolean
value = !!(string =~ /t|true|y|yes|1/i)
when :string
value = string
when :time
begin
value = Time.parse string
rescue ArgumentError => e
invalid_format string, e.message
end
when :date
begin
value = Date.parse string
rescue ArgumentError => e
invalid_format string, e.message
end
when :datetime
begin
value = DateTime.parse string
rescue ArgumentError => e
invalid_format string, e.message
end
when :double
if string =~ /-?\d*(\.\d+)/
value = Float(string)
elsif string =~ /^(-?)Inf(inity)?$/i
if $1
value = -Float::INFINITY
else
value = Float::INFINITY
end
2017-08-02 01:42:49 -04:00
elsif string =~ /^(-?)NaN$/i
if $1
value = -Float::NAN
else
value = Float::NAN
2015-07-14 19:01:38 -04:00
end
2017-08-02 01:42:49 -04:00
else
invalid_format string
end
when :category_id
if string =~ /(.*)\/(.*)/
parent_name = $1
child_name = $2
parent = Category.query_parent_category(parent_name)
invalid_format string, "Could not find category named #{parent_name}" unless parent
object = Category.query_category(child_name, parent)
invalid_format string, "Could not find subcategory of #{parent_name} named #{child_name}" unless object
else
object = Category.where(id: string.to_i).first || Category.where(slug: string).first || Category.where(name: string).first
invalid_format string, "Could not find category named #{string}" unless object
end
value = object.id
when :user_id, :post_id, :topic_id, :group_id, :badge_id
if string.gsub(/[ _]/, '') =~ /^-?\d+$/
clazz_name = (/^(.*)_id$/.match(type.to_s)[1].classify.to_sym)
2015-07-14 19:01:38 -04:00
begin
2017-08-02 01:42:49 -04:00
object = Object.const_get(clazz_name).with_deleted.find(string.gsub(/[ _]/, '').to_i)
value = object.id
rescue ActiveRecord::RecordNotFound
invalid_format string, "The specified #{clazz_name} was not found"
2015-07-14 19:01:38 -04:00
end
2017-08-02 01:42:49 -04:00
elsif type == :user_id
begin
object = User.find_by_username_or_email(string)
value = object.id
rescue ActiveRecord::RecordNotFound
invalid_format string, "The user named #{string} was not found"
2015-07-15 15:23:56 -04:00
end
2017-08-02 01:42:49 -04:00
elsif type == :post_id
if string =~ /(\d+)\/(\d+)(\?u=.*)?$/
object = Post.with_deleted.find_by(topic_id: $1, post_number: $2)
invalid_format string, "The post at topic:#{$1} post_number:#{$2} was not found" unless object
value = object.id
end
2017-08-02 01:42:49 -04:00
elsif type == :topic_id
if string =~ /\/t\/[^\/]+\/(\d+)/
begin
2017-08-02 01:42:49 -04:00
object = Topic.with_deleted.find($1)
2015-08-04 02:35:02 -04:00
value = object.id
rescue ActiveRecord::RecordNotFound
2017-08-02 01:42:49 -04:00
invalid_format string, "The topic with id #{$1} was not found"
end
end
2017-08-02 01:42:49 -04:00
elsif type == :group_id
object = Group.where(name: string).first
invalid_format string, "The group named #{string} was not found" unless object
value = object.id
else
2017-08-02 01:42:49 -04:00
invalid_format string
end
when :int_list
value = string.split(',').map { |s| s.downcase == '#null' ? nil : s.to_i }
invalid_format string, "can't be empty" if value.length == 0
when :string_list
value = string.split(',').map { |s| s.downcase == '#null' ? nil : s }
invalid_format string, "can't be empty" if value.length == 0
when :user_list
value = string.split(',').map { |s| User.find_by_username_or_email(s) }
invalid_format string, "can't be empty" if value.length == 0
else
raise TypeError.new('unknown parameter type??? should not get here')
end
value
end
2017-08-02 01:42:49 -04:00
def self.create_from_sql(sql, opts = {})
in_params = false
ret_params = []
sql.lines.find do |line|
line.chomp!
if in_params
# -- (ident) :(ident) (= (ident))?
if line =~ /^\s*--\s*([a-zA-Z_ ]+)\s*:([a-z_]+)\s*(?:=\s+(.*)\s*)?$/
type = $1
ident = $2
default = $3
nullable = false
if type =~ /^(null)?(.*?)(null)?$/i
2017-08-02 01:42:49 -04:00
if $1 || $3
nullable = true
end
type = $2
end
type = type.strip
2015-07-14 19:01:38 -04:00
begin
ret_params << DataExplorer::Parameter.new(ident, type, default, nullable)
rescue
if opts[:strict]
raise
end
end
false
elsif line =~ /^\s+$/
false
else
true
end
else
if line =~ /^\s*--\s*\[params\]\s*$/
in_params = true
end
false
end
end
return ret_params
end
end
require_dependency 'application_controller'
require_dependency File.expand_path('../lib/queries.rb', __FILE__)
2015-06-25 16:26:31 -04:00
class DataExplorer::QueryController < ::ApplicationController
requires_plugin DataExplorer.plugin_name
before_action :check_enabled
2015-07-08 16:45:13 -04:00
def check_enabled
raise Discourse::NotFound unless SiteSetting.data_explorer_enabled?
end
def index
# guardian.ensure_can_use_data_explorer!
queries = DataExplorer::Query.all
Queries.default.each do |params|
query = DataExplorer::Query.new
query.id = params.second["id"]
query.sql = params.second["sql"]
query.name = params.second["name"]
query.description = params.second["description"]
query.created_by = Discourse::SYSTEM_USER_ID.to_s
# don't render this query if query with the same id already exists in pstore
queries.push(query) unless DataExplorer.pstore_get("q:#{query.id}").present?
end
2015-06-25 16:26:31 -04:00
render_serialized queries, DataExplorer::QuerySerializer, root: 'queries'
end
skip_before_action :check_xhr, only: [:show]
def show
2015-06-30 15:52:17 -04:00
check_xhr unless params[:export]
query = DataExplorer::Query.find(params[:id].to_i)
if params[:export]
2015-06-30 15:52:17 -04:00
response.headers['Content-Disposition'] = "attachment; filename=#{query.slug}.dcquery.json"
response.sending_file = true
end
# guardian.ensure_can_see! query
2015-06-30 13:37:48 -04:00
render_serialized query, DataExplorer::QuerySerializer, root: 'query'
end
def create
# guardian.ensure_can_create_explorer_query!
2015-06-29 15:10:24 -04:00
query = DataExplorer::Query.from_hash params.require(:query)
query.created_at = Time.now
query.created_by = current_user.id.to_s
query.last_run_at = Time.now
2015-06-30 15:52:17 -04:00
query.id = nil # json import will assign an id, which is wrong
query.save
2015-06-30 15:52:17 -04:00
render_serialized query, DataExplorer::QuerySerializer, root: 'query'
end
def update
2015-06-30 15:52:17 -04:00
query = DataExplorer::Query.find(params[:id].to_i, ignore_deleted: true)
2015-06-29 15:10:24 -04:00
hash = params.require(:query)
2015-06-30 15:52:17 -04:00
# Undeleting
unless query.id
if hash[:id]
query.id = hash[:id].to_i
else
raise Discourse::NotFound
end
end
[:name, :sql, :description].each do |sym|
2015-06-30 13:20:22 -04:00
query.send("#{sym}=", hash[sym]) if hash[sym]
end
2015-07-14 19:01:38 -04:00
query.check_params!
query.save
render_serialized query, DataExplorer::QuerySerializer, root: 'query'
rescue DataExplorer::ValidationError => e
render_json_error e.message
end
def destroy
query = DataExplorer::Query.find(params[:id].to_i)
query.destroy
2015-06-30 15:52:17 -04:00
2017-08-02 01:42:49 -04:00
render json: { success: true, errors: [] }
end
2015-07-08 16:45:13 -04:00
def schema
schema_version = DB.query_single("SELECT max(version) AS tag FROM schema_migrations").first
if stale?(public: true, etag: schema_version, template: false)
2015-07-08 16:45:13 -04:00
render json: DataExplorer.schema
end
end
skip_before_action :check_xhr, only: [:run]
2015-06-30 22:51:38 -04:00
# Return value:
# success - true/false. if false, inspect the errors value.
# errors - array of strings.
# params - hash. Echo of the query parameters as executed.
# duration - float. Time to execute the query, in milliseconds, to 1 decimal place.
# columns - array of strings. Titles of the returned columns, in order.
# explain - string. (Optional - pass explain=true in the request) Postgres query plan, UNIX newlines.
# rows - array of array of strings. Results of the query. In the same order as 'columns'.
def run
check_xhr unless params[:download]
query = DataExplorer::Query.find(params[:id].to_i)
query.last_run_at = Time.now
if params[:id].to_i < 0
query.created_by = Discourse::SYSTEM_USER_ID.to_s
query.save_default_query
else
query.save
end
if params[:download]
response.sending_file = true
end
2015-07-15 15:23:56 -04:00
params[:params] = params[:_params] if params[:_params] # testing workaround
query_params = {}
query_params = MultiJson.load(params[:params]) if params[:params]
opts = { current_user: current_user.username }
opts[:explain] = true if params[:explain] == "true"
opts[:limit] = "ALL" if params[:format] == "csv"
2015-06-30 22:51:38 -04:00
opts[:limit] = params[:limit].to_i if params[:limit]
result = DataExplorer.run_query(query, query_params, opts)
if result[:error]
err = result[:error]
# Pretty printing logic
err_class = err.class
err_msg = err.message
if err.is_a? ActiveRecord::StatementInvalid
err_class = err.original_exception.class
err_msg.gsub!("#{err_class}:", '')
else
err_msg = "#{err_class}: #{err_msg}"
end
render json: {
success: false,
errors: [err_msg]
2015-07-15 15:51:33 -04:00
}, status: 422
else
pg_result = result[:pg_result]
cols = pg_result.fields
2015-08-03 18:07:29 -04:00
respond_to do |format|
format.json do
if params[:download]
response.headers['Content-Disposition'] =
"attachment; filename=#{query.slug}@#{Slug.for(Discourse.current_hostname, 'discourse')}-#{Date.today}.dcqresult.json"
end
json = {
success: true,
errors: [],
duration: (result[:duration_secs].to_f * 1000).round(1),
result_count: pg_result.values.length || 0,
2015-08-03 18:07:29 -04:00
params: query_params,
columns: cols,
result_limit: DataExplorer::QUERY_RESULT_MAX_LIMIT
2015-08-03 18:07:29 -04:00
}
json[:explain] = result[:explain] if opts[:explain]
2015-08-25 23:48:19 -04:00
ext = DataExplorer.add_extra_data(pg_result)
json[:colrender] = ext[1]
json[:relations] = ext[0]
2015-08-03 18:07:29 -04:00
json[:rows] = pg_result.values
render json: json
end
format.csv do
response.headers['Content-Disposition'] =
"attachment; filename=#{query.slug}@#{Slug.for(Discourse.current_hostname, 'discourse')}-#{Date.today}.dcqresult.csv"
require 'csv'
text = CSV.generate do |csv|
csv << cols
pg_result.values.each do |row|
csv << row
end
end
render plain: text
2015-08-03 18:07:29 -04:00
end
end
end
end
end
class DataExplorer::QuerySerializer < ActiveModel::Serializer
attributes :id, :sql, :name, :description, :param_info, :created_by, :created_at, :username, :last_run_at
def param_info
object.params.map(&:to_hash) rescue nil
end
def username
User.find(created_by).username rescue nil
end
end
DataExplorer::Engine.routes.draw do
2015-06-25 16:26:31 -04:00
root to: "query#index"
2015-07-08 16:45:13 -04:00
get 'schema' => "query#schema"
2015-06-25 16:26:31 -04:00
get 'queries' => "query#index"
2015-06-29 15:10:24 -04:00
post 'queries' => "query#create"
2015-06-30 13:20:22 -04:00
get 'queries/:id' => "query#show"
put 'queries/:id' => "query#update"
delete 'queries/:id' => "query#destroy"
post 'queries/:id/run' => "query#run"
end
Discourse::Application.routes.append do
2015-06-25 16:26:31 -04:00
mount ::DataExplorer::Engine, at: '/admin/plugins/explorer', constraints: AdminConstraint.new
end
end