2019-05-02 18:17:27 -04:00
|
|
|
# frozen_string_literal: true
|
|
|
|
|
2018-05-22 15:42:49 -04:00
|
|
|
require 'sqlite3'
|
|
|
|
|
|
|
|
module ImportScripts
|
|
|
|
class GenericDatabase
|
2019-02-28 15:59:36 -05:00
|
|
|
def initialize(directory, batch_size:, recreate: false, numeric_keys: false)
|
2018-05-22 15:42:49 -04:00
|
|
|
filename = "#{directory}/index.db"
|
2022-01-05 12:45:08 -05:00
|
|
|
File.delete(filename) if recreate && File.exist?(filename)
|
2018-05-22 15:42:49 -04:00
|
|
|
|
|
|
|
@db = SQLite3::Database.new(filename, results_as_hash: true)
|
|
|
|
@batch_size = batch_size
|
2019-02-28 15:59:36 -05:00
|
|
|
@numeric_keys = numeric_keys
|
2018-05-22 15:42:49 -04:00
|
|
|
|
|
|
|
configure_database
|
|
|
|
create_category_table
|
2019-08-13 11:24:22 -04:00
|
|
|
create_like_table
|
2018-05-22 15:42:49 -04:00
|
|
|
create_user_table
|
|
|
|
create_topic_table
|
|
|
|
create_post_table
|
|
|
|
end
|
|
|
|
|
|
|
|
def insert_category(category)
|
|
|
|
@db.execute(<<-SQL, prepare(category))
|
|
|
|
INSERT OR REPLACE INTO category (id, name, description, position, url)
|
|
|
|
VALUES (:id, :name, :description, :position, :url)
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
|
|
|
def insert_user(user)
|
|
|
|
@db.execute(<<-SQL, prepare(user))
|
2019-02-28 15:59:36 -05:00
|
|
|
INSERT OR REPLACE
|
|
|
|
INTO user (id, email, username, name, bio, avatar_path, created_at, last_seen_at, active)
|
|
|
|
VALUES (:id, :email, :username, :name, :bio, :avatar_path, :created_at, :last_seen_at, :active)
|
2018-05-22 15:42:49 -04:00
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
2019-08-13 11:24:22 -04:00
|
|
|
def insert_like(like)
|
|
|
|
@db.execute(<<-SQL, prepare(like))
|
|
|
|
INSERT OR REPLACE INTO like (id, user_id, post_id, topic)
|
|
|
|
VALUES (:id, :user_id, :post_id, :topic)
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
2018-05-22 15:42:49 -04:00
|
|
|
def insert_topic(topic)
|
2019-08-13 11:24:22 -04:00
|
|
|
like_user_ids = topic.delete(:like_user_ids)
|
2019-02-28 15:59:36 -05:00
|
|
|
attachments = topic.delete(:attachments)
|
|
|
|
topic[:upload_count] = attachments&.size || 0
|
|
|
|
|
2019-08-13 11:24:22 -04:00
|
|
|
@db.transaction do
|
|
|
|
@db.execute(<<-SQL, prepare(topic))
|
|
|
|
INSERT OR REPLACE INTO topic (id, title, raw, category_id, closed, user_id, created_at, url, upload_count)
|
|
|
|
VALUES (:id, :title, :raw, :category_id, :closed, :user_id, :created_at, :url, :upload_count)
|
2019-02-28 15:59:36 -05:00
|
|
|
SQL
|
2019-08-13 11:24:22 -04:00
|
|
|
|
|
|
|
attachments&.each do |attachment|
|
|
|
|
@db.execute(<<-SQL, topic_id: topic[:id], path: attachment)
|
|
|
|
INSERT OR REPLACE INTO topic_upload (topic_id, path)
|
|
|
|
VALUES (:topic_id, :path)
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
|
|
|
like_user_ids&.each do |user_id|
|
|
|
|
@db.execute(<<-SQL, topic_id: topic[:id], user_id: user_id)
|
|
|
|
INSERT OR REPLACE INTO like (topic_id, user_id)
|
|
|
|
VALUES (:topic_id, :user_id)
|
|
|
|
SQL
|
|
|
|
end
|
2019-02-28 15:59:36 -05:00
|
|
|
end
|
2018-05-22 15:42:49 -04:00
|
|
|
end
|
|
|
|
|
|
|
|
def insert_post(post)
|
2019-08-13 11:24:22 -04:00
|
|
|
like_user_ids = post.delete(:like_user_ids)
|
2019-02-28 15:59:36 -05:00
|
|
|
attachments = post.delete(:attachments)
|
|
|
|
post[:upload_count] = attachments&.size || 0
|
|
|
|
|
2019-08-13 11:24:22 -04:00
|
|
|
@db.transaction do
|
|
|
|
@db.execute(<<-SQL, prepare(post))
|
|
|
|
INSERT OR REPLACE INTO post (id, raw, topic_id, user_id, created_at, reply_to_post_id, url, upload_count)
|
|
|
|
VALUES (:id, :raw, :topic_id, :user_id, :created_at, :reply_to_post_id, :url, :upload_count)
|
2019-02-28 15:59:36 -05:00
|
|
|
SQL
|
2019-08-13 11:24:22 -04:00
|
|
|
|
|
|
|
attachments&.each do |attachment|
|
|
|
|
@db.execute(<<-SQL, post_id: post[:id], path: attachment)
|
|
|
|
INSERT OR REPLACE INTO post_upload (post_id, path)
|
|
|
|
VALUES (:post_id, :path)
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
|
|
|
like_user_ids&.each do |user_id|
|
|
|
|
@db.execute(<<-SQL, post_id: post[:id], user_id: user_id)
|
|
|
|
INSERT OR REPLACE INTO like (post_id, user_id)
|
|
|
|
VALUES (:post_id, :user_id)
|
|
|
|
SQL
|
|
|
|
end
|
2019-02-28 15:59:36 -05:00
|
|
|
end
|
2018-05-22 15:42:49 -04:00
|
|
|
end
|
|
|
|
|
|
|
|
def sort_posts_by_created_at
|
|
|
|
@db.execute 'DELETE FROM post_order'
|
|
|
|
|
|
|
|
@db.execute <<-SQL
|
2019-02-28 15:59:36 -05:00
|
|
|
INSERT INTO post_order (post_id)
|
2018-05-22 15:42:49 -04:00
|
|
|
SELECT id
|
|
|
|
FROM post
|
|
|
|
ORDER BY created_at, topic_id, id
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
2019-02-28 15:59:36 -05:00
|
|
|
def delete_unused_users
|
|
|
|
@db.execute <<~SQL
|
|
|
|
DELETE FROM user
|
|
|
|
WHERE NOT EXISTS(
|
|
|
|
SELECT 1
|
|
|
|
FROM topic
|
|
|
|
WHERE topic.user_id = user.id
|
|
|
|
) AND NOT EXISTS(
|
|
|
|
SELECT 1
|
|
|
|
FROM post
|
|
|
|
WHERE post.user_id = user.id
|
|
|
|
)
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
2018-05-22 15:42:49 -04:00
|
|
|
def fetch_categories
|
|
|
|
@db.execute(<<-SQL)
|
|
|
|
SELECT *
|
|
|
|
FROM category
|
|
|
|
ORDER BY position, name
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
|
|
|
def count_users
|
|
|
|
@db.get_first_value(<<-SQL)
|
|
|
|
SELECT COUNT(*)
|
|
|
|
FROM user
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
|
|
|
def fetch_users(last_id)
|
|
|
|
rows = @db.execute(<<-SQL, last_id)
|
|
|
|
SELECT *
|
|
|
|
FROM user
|
|
|
|
WHERE id > :last_id
|
|
|
|
ORDER BY id
|
|
|
|
LIMIT #{@batch_size}
|
|
|
|
SQL
|
|
|
|
|
|
|
|
add_last_column_value(rows, 'id')
|
|
|
|
end
|
|
|
|
|
2019-02-28 15:59:36 -05:00
|
|
|
def get_user_id(username)
|
|
|
|
@db.get_first_value(<<-SQL, username)
|
|
|
|
SELECT id
|
|
|
|
FROM user
|
|
|
|
WHERE username = :username
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
2018-05-22 15:42:49 -04:00
|
|
|
def count_topics
|
|
|
|
@db.get_first_value(<<-SQL)
|
|
|
|
SELECT COUNT(*)
|
|
|
|
FROM topic
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
|
|
|
def fetch_topics(last_id)
|
|
|
|
rows = @db.execute(<<-SQL, last_id)
|
|
|
|
SELECT *
|
|
|
|
FROM topic
|
|
|
|
WHERE id > :last_id
|
|
|
|
ORDER BY id
|
|
|
|
LIMIT #{@batch_size}
|
|
|
|
SQL
|
|
|
|
|
|
|
|
add_last_column_value(rows, 'id')
|
|
|
|
end
|
|
|
|
|
2019-02-28 15:59:36 -05:00
|
|
|
def fetch_topic_attachments(topic_id)
|
|
|
|
@db.execute(<<-SQL, topic_id)
|
|
|
|
SELECT path
|
|
|
|
FROM topic_upload
|
|
|
|
WHERE topic_id = :topic_id
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
2018-05-22 15:42:49 -04:00
|
|
|
def count_posts
|
|
|
|
@db.get_first_value(<<-SQL)
|
|
|
|
SELECT COUNT(*)
|
|
|
|
FROM post
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
|
|
|
def fetch_posts(last_row_id)
|
|
|
|
rows = @db.execute(<<-SQL, last_row_id)
|
2019-02-28 15:59:36 -05:00
|
|
|
SELECT ROWID AS rowid, *
|
|
|
|
FROM post
|
|
|
|
WHERE ROWID > :last_row_id
|
|
|
|
ORDER BY ROWID
|
|
|
|
LIMIT #{@batch_size}
|
|
|
|
SQL
|
|
|
|
|
|
|
|
add_last_column_value(rows, 'rowid')
|
|
|
|
end
|
|
|
|
|
|
|
|
def fetch_sorted_posts(last_row_id)
|
|
|
|
rows = @db.execute(<<-SQL, last_row_id)
|
|
|
|
SELECT o.ROWID AS rowid, p.*
|
2018-05-22 15:42:49 -04:00
|
|
|
FROM post p
|
2019-02-28 15:59:36 -05:00
|
|
|
JOIN post_order o ON (p.id = o.post_id)
|
2018-05-22 15:42:49 -04:00
|
|
|
WHERE o.ROWID > :last_row_id
|
|
|
|
ORDER BY o.ROWID
|
|
|
|
LIMIT #{@batch_size}
|
|
|
|
SQL
|
|
|
|
|
|
|
|
add_last_column_value(rows, 'rowid')
|
|
|
|
end
|
|
|
|
|
2019-02-28 15:59:36 -05:00
|
|
|
def fetch_post_attachments(post_id)
|
|
|
|
@db.execute(<<-SQL, post_id)
|
|
|
|
SELECT path
|
|
|
|
FROM post_upload
|
|
|
|
WHERE post_id = :post_id
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
2019-08-13 11:24:22 -04:00
|
|
|
def count_likes
|
|
|
|
@db.get_first_value(<<-SQL)
|
|
|
|
SELECT COUNT(*)
|
|
|
|
FROM like
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
|
|
|
def fetch_likes(last_row_id)
|
|
|
|
rows = @db.execute(<<-SQL, last_row_id)
|
|
|
|
SELECT ROWID AS rowid, *
|
|
|
|
FROM like
|
|
|
|
WHERE ROWID > :last_row_id
|
|
|
|
ORDER BY ROWID
|
|
|
|
LIMIT #{@batch_size}
|
|
|
|
SQL
|
|
|
|
|
|
|
|
add_last_column_value(rows, 'rowid')
|
|
|
|
end
|
|
|
|
|
2018-05-22 15:42:49 -04:00
|
|
|
def execute_sql(sql)
|
|
|
|
@db.execute(sql)
|
|
|
|
end
|
|
|
|
|
2018-08-28 10:21:39 -04:00
|
|
|
def get_first_value(sql)
|
|
|
|
@db.get_first_value(sql)
|
|
|
|
end
|
|
|
|
|
2018-05-22 15:42:49 -04:00
|
|
|
private
|
|
|
|
|
|
|
|
def configure_database
|
|
|
|
@db.execute 'PRAGMA journal_mode = OFF'
|
|
|
|
@db.execute 'PRAGMA locking_mode = EXCLUSIVE'
|
|
|
|
end
|
|
|
|
|
2019-02-28 15:59:36 -05:00
|
|
|
def key_data_type
|
|
|
|
@numeric_keys ? 'INTEGER' : 'TEXT'
|
|
|
|
end
|
|
|
|
|
2018-05-22 15:42:49 -04:00
|
|
|
def create_category_table
|
|
|
|
@db.execute <<-SQL
|
|
|
|
CREATE TABLE IF NOT EXISTS category (
|
2019-02-28 15:59:36 -05:00
|
|
|
id #{key_data_type} NOT NULL PRIMARY KEY,
|
2018-05-22 15:42:49 -04:00
|
|
|
name TEXT NOT NULL,
|
|
|
|
description TEXT,
|
|
|
|
position INTEGER,
|
|
|
|
url TEXT
|
|
|
|
)
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
2019-08-13 11:24:22 -04:00
|
|
|
def create_like_table
|
|
|
|
@db.execute <<-SQL
|
|
|
|
CREATE TABLE IF NOT EXISTS like (
|
|
|
|
user_id #{key_data_type} NOT NULL,
|
|
|
|
topic_id #{key_data_type},
|
|
|
|
post_id #{key_data_type}
|
|
|
|
)
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
2018-05-22 15:42:49 -04:00
|
|
|
def create_user_table
|
|
|
|
@db.execute <<-SQL
|
|
|
|
CREATE TABLE IF NOT EXISTS user (
|
2019-02-28 15:59:36 -05:00
|
|
|
id #{key_data_type} NOT NULL PRIMARY KEY,
|
2018-05-22 15:42:49 -04:00
|
|
|
email TEXT,
|
|
|
|
username TEXT,
|
|
|
|
name TEXT,
|
2019-02-28 15:59:36 -05:00
|
|
|
bio TEXT,
|
|
|
|
avatar_path TEXT,
|
2018-05-22 15:42:49 -04:00
|
|
|
created_at DATETIME,
|
|
|
|
last_seen_at DATETIME,
|
|
|
|
active BOOLEAN NOT NULL DEFAULT true
|
|
|
|
)
|
|
|
|
SQL
|
2019-02-28 15:59:36 -05:00
|
|
|
|
|
|
|
@db.execute 'CREATE INDEX IF NOT EXISTS user_by_username ON user (username)'
|
2018-05-22 15:42:49 -04:00
|
|
|
end
|
|
|
|
|
|
|
|
def create_topic_table
|
|
|
|
@db.execute <<-SQL
|
|
|
|
CREATE TABLE IF NOT EXISTS topic (
|
2019-02-28 15:59:36 -05:00
|
|
|
id #{key_data_type} NOT NULL PRIMARY KEY,
|
2018-05-22 15:42:49 -04:00
|
|
|
title TEXT,
|
|
|
|
raw TEXT,
|
2019-02-28 15:59:36 -05:00
|
|
|
category_id #{key_data_type},
|
2018-05-22 15:42:49 -04:00
|
|
|
closed BOOLEAN NOT NULL DEFAULT false,
|
2019-02-28 15:59:36 -05:00
|
|
|
user_id #{key_data_type} NOT NULL,
|
2018-05-22 15:42:49 -04:00
|
|
|
created_at DATETIME,
|
2019-02-28 15:59:36 -05:00
|
|
|
url TEXT,
|
|
|
|
upload_count INTEGER DEFAULT 0
|
2018-05-22 15:42:49 -04:00
|
|
|
)
|
|
|
|
SQL
|
|
|
|
|
|
|
|
@db.execute 'CREATE INDEX IF NOT EXISTS topic_by_user_id ON topic (user_id)'
|
2019-02-28 15:59:36 -05:00
|
|
|
|
|
|
|
@db.execute <<-SQL
|
|
|
|
CREATE TABLE IF NOT EXISTS topic_upload (
|
|
|
|
topic_id #{key_data_type} NOT NULL,
|
|
|
|
path TEXT NOT NULL
|
|
|
|
)
|
|
|
|
SQL
|
|
|
|
|
|
|
|
@db.execute 'CREATE UNIQUE INDEX IF NOT EXISTS topic_upload_unique ON topic_upload(topic_id, path)'
|
2018-05-22 15:42:49 -04:00
|
|
|
end
|
|
|
|
|
|
|
|
def create_post_table
|
|
|
|
@db.execute <<-SQL
|
|
|
|
CREATE TABLE IF NOT EXISTS post (
|
2019-02-28 15:59:36 -05:00
|
|
|
id #{key_data_type} NOT NULL PRIMARY KEY,
|
2018-05-22 15:42:49 -04:00
|
|
|
raw TEXT,
|
2019-02-28 15:59:36 -05:00
|
|
|
topic_id #{key_data_type} NOT NULL,
|
|
|
|
user_id #{key_data_type} NOT NULL,
|
2018-05-22 15:42:49 -04:00
|
|
|
created_at DATETIME,
|
2019-02-28 15:59:36 -05:00
|
|
|
reply_to_post_id #{key_data_type},
|
|
|
|
url TEXT,
|
|
|
|
upload_count INTEGER DEFAULT 0
|
2018-05-22 15:42:49 -04:00
|
|
|
)
|
|
|
|
SQL
|
|
|
|
|
|
|
|
@db.execute 'CREATE INDEX IF NOT EXISTS post_by_user_id ON post (user_id)'
|
|
|
|
|
|
|
|
@db.execute <<-SQL
|
|
|
|
CREATE TABLE IF NOT EXISTS post_order (
|
2019-02-28 15:59:36 -05:00
|
|
|
post_id #{key_data_type} NOT NULL PRIMARY KEY
|
2018-05-22 15:42:49 -04:00
|
|
|
)
|
|
|
|
SQL
|
2019-02-28 15:59:36 -05:00
|
|
|
|
|
|
|
@db.execute <<-SQL
|
|
|
|
CREATE TABLE IF NOT EXISTS post_upload (
|
|
|
|
post_id #{key_data_type} NOT NULL,
|
|
|
|
path TEXT NOT NULL
|
|
|
|
)
|
|
|
|
SQL
|
|
|
|
|
|
|
|
@db.execute 'CREATE UNIQUE INDEX IF NOT EXISTS post_upload_unique ON post_upload(post_id, path)'
|
2018-05-22 15:42:49 -04:00
|
|
|
end
|
|
|
|
|
|
|
|
def prepare(hash)
|
|
|
|
hash.each do |key, value|
|
|
|
|
if value.is_a?(TrueClass) || value.is_a?(FalseClass)
|
|
|
|
hash[key] = value ? 1 : 0
|
|
|
|
elsif value.is_a?(Date)
|
|
|
|
hash[key] = value.to_s
|
|
|
|
end
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
def add_last_column_value(rows, *last_columns)
|
|
|
|
return rows if last_columns.empty?
|
|
|
|
|
|
|
|
result = [rows]
|
|
|
|
last_row = rows.last
|
|
|
|
|
|
|
|
last_columns.each { |column| result.push(last_row ? last_row[column] : nil) }
|
|
|
|
result
|
|
|
|
end
|
|
|
|
end
|
|
|
|
end
|