|
| 1 | +require "json" |
| 2 | +require "net/http" |
| 3 | +require "uri" |
| 4 | +require "sqlite3" |
| 5 | +require "mini_sql" |
| 6 | +require "cgi" |
| 7 | + |
| 8 | +begin |
| 9 | + config = JSON.parse(File.read("config.json")) |
| 10 | +rescue StandardError |
| 11 | + puts "Please create a file called .creds with your API KEY and USERNAME" |
| 12 | +end |
| 13 | + |
| 14 | +# Replace these values with your Discourse instance details |
| 15 | +DISCOURSE_DOMAIN = config["domain"] |
| 16 | +API_KEY = config["api_key"] |
| 17 | +API_USERNAME = config["api_username"] |
| 18 | +TOPIC_QUERY_ID = config["topics_query_id"] |
| 19 | +POST_QUERY_ID = config["posts_query_id"] |
| 20 | + |
| 21 | +sqlite_conn = SQLite3::Database.new("dump.db") |
| 22 | +conn = MiniSql::Connection.get(sqlite_conn) |
| 23 | + |
| 24 | +def run_report(query_id:, min_id: 0, limit:) |
| 25 | + params = CGI.escape({ min_id: min_id.to_s }.to_json) |
| 26 | + |
| 27 | + uri = |
| 28 | + URI( |
| 29 | + "https://#{DISCOURSE_DOMAIN}/admin/plugins/explorer/queries/#{query_id}/run?limit=#{limit}¶ms=#{params}" |
| 30 | + ) |
| 31 | + http = Net::HTTP.new(uri.host, uri.port) |
| 32 | + http.use_ssl = true |
| 33 | + |
| 34 | + request = Net::HTTP::Post.new(uri.request_uri) |
| 35 | + request["Content-Type"] = "application/json" |
| 36 | + request["Api-Key"] = API_KEY |
| 37 | + request["Api-Username"] = API_USERNAME |
| 38 | + |
| 39 | + response = http.request(request) |
| 40 | + JSON.parse(response.body) |
| 41 | +end |
| 42 | + |
| 43 | +def create_schema(conn) |
| 44 | + conn.exec <<-SQL |
| 45 | + CREATE TABLE IF NOT EXISTS topics ( |
| 46 | + id INTEGER PRIMARY KEY, |
| 47 | + category, |
| 48 | + title, |
| 49 | + created_at, |
| 50 | + user_id, |
| 51 | + tags |
| 52 | + ) |
| 53 | + SQL |
| 54 | + |
| 55 | + conn.exec <<-SQL |
| 56 | + CREATE TABLE IF NOT EXISTS users( |
| 57 | + id INTEGER PRIMARY KEY, |
| 58 | + username, |
| 59 | + name |
| 60 | + ) |
| 61 | + SQL |
| 62 | + |
| 63 | + conn.exec <<-SQL |
| 64 | + CREATE TABLE IF NOT EXISTS posts( |
| 65 | + id INTEGER PRIMARY KEY, |
| 66 | + raw, |
| 67 | + post_number, |
| 68 | + topic_id, |
| 69 | + user_id, |
| 70 | + created_at |
| 71 | + ) |
| 72 | + SQL |
| 73 | +end |
| 74 | + |
| 75 | +def load_posts(conn, rows) |
| 76 | + highest_id = 0 |
| 77 | + posts_loaded = 0 |
| 78 | + |
| 79 | + conn.exec "BEGIN TRANSACTION" |
| 80 | + |
| 81 | + rows.each do |row| |
| 82 | + conn.exec <<~SQL, *row |
| 83 | + INSERT OR IGNORE INTO posts (id, raw, post_number, topic_id, user_id, created_at) |
| 84 | + VALUES (?, ?, ?, ?, ?, ?) |
| 85 | + SQL |
| 86 | + posts_loaded += 1 |
| 87 | + highest_id = row[0] if row[0] > highest_id |
| 88 | + end |
| 89 | + |
| 90 | + conn.exec "COMMIT TRANSACTION" |
| 91 | + |
| 92 | + { highest_id: highest_id, posts_loaded: posts_loaded } |
| 93 | +end |
| 94 | + |
| 95 | +def load_topics(conn, rows) |
| 96 | + highest_id = 0 |
| 97 | + topics_loaded = 0 |
| 98 | + |
| 99 | + conn.exec "BEGIN TRANSACTION" |
| 100 | + |
| 101 | + rows.each do |row| |
| 102 | + conn.exec <<~SQL, *row |
| 103 | + INSERT OR IGNORE INTO topics (id, category, title, created_at, user_id, tags) |
| 104 | + VALUES (?, ?, ?, ?, ?, ?) |
| 105 | + SQL |
| 106 | + topics_loaded += 1 |
| 107 | + highest_id = row[0] if row[0] > highest_id |
| 108 | + end |
| 109 | + |
| 110 | + conn.exec "COMMIT TRANSACTION" |
| 111 | + |
| 112 | + { highest_id: highest_id, topics_loaded: topics_loaded } |
| 113 | +end |
| 114 | + |
| 115 | +def load_users(conn, rows) |
| 116 | + conn.exec "BEGIN TRANSACTION" |
| 117 | + loaded = 0 |
| 118 | + |
| 119 | + rows.each do |row| |
| 120 | + conn.exec <<~SQL, *row |
| 121 | + INSERT OR IGNORE INTO users(id, username, name) |
| 122 | + VALUES (?, ?, ?) |
| 123 | + SQL |
| 124 | + loaded += 1 |
| 125 | + end |
| 126 | + |
| 127 | + conn.exec "COMMIT TRANSACTION" |
| 128 | + loaded |
| 129 | +end |
| 130 | + |
| 131 | +def load_users_from_json(conn, json) |
| 132 | + users = json.dig("relations", "user") |
| 133 | + if users |
| 134 | + users = users.map { |user| [user["id"], user["username"], user["name"]] } |
| 135 | + loaded = load_users(conn, users) |
| 136 | + puts "Loaded #{loaded} users" |
| 137 | + end |
| 138 | +end |
| 139 | + |
| 140 | +create_schema(conn) |
| 141 | + |
| 142 | +min_id = 0 |
| 143 | +while true |
| 144 | + response_data = |
| 145 | + run_report(query_id: TOPIC_QUERY_ID, min_id: min_id, limit: 10_000) |
| 146 | + |
| 147 | + load_users_from_json(conn, response_data) |
| 148 | + |
| 149 | + result = load_topics(conn, response_data["rows"]) |
| 150 | + puts "Loaded #{result[:topics_loaded]} topics (highest id is #{result[:highest_id]})" |
| 151 | + |
| 152 | + min_id = result[:highest_id] |
| 153 | + break if result[:topics_loaded] == 0 |
| 154 | +end |
| 155 | + |
| 156 | +min_id = 0 |
| 157 | +while true |
| 158 | + response_data = |
| 159 | + run_report(query_id: POST_QUERY_ID, min_id: min_id, limit: 10_000) |
| 160 | + |
| 161 | + load_users_from_json(conn, response_data) |
| 162 | + |
| 163 | + result = load_posts(conn, response_data["rows"]) |
| 164 | + puts "Loaded #{result[:posts_loaded]} posts (highest id is #{result[:highest_id]})" |
| 165 | + |
| 166 | + min_id = result[:highest_id] |
| 167 | + break if result[:posts_loaded] == 0 |
| 168 | +end |
0 commit comments