discourse/spec/requests/session_controller_spec.rb

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

2973 lines
106 KiB
Ruby
Raw Normal View History

# frozen_string_literal: true
require "rotp"
RSpec.describe SessionController do
let(:user) { Fabricate(:user) }
let(:email_token) { Fabricate(:email_token, user: user) }
fab!(:admin) { Fabricate(:admin) }
let(:admin_email_token) { Fabricate(:email_token, user: admin) }
shared_examples "failed to continue local login" do
it "should return the right response" do
expect(response).not_to be_successful
expect(response.status).to eq(403)
end
end
describe "#email_login_info" do
let(:email_token) do
Fabricate(:email_token, user: user, scope: EmailToken.scopes[:email_login])
end
before { SiteSetting.enable_local_logins_via_email = true }
context "when local logins via email disabled" do
before { SiteSetting.enable_local_logins_via_email = false }
it "only works for admins" do
get "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(403)
user.update(admin: true)
get "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
end
end
context "when SSO enabled" do
before do
FEATURE: Rename 'Discourse SSO' to DiscourseConnect (#11978) The 'Discourse SSO' protocol is being rebranded to DiscourseConnect. This should help to reduce confusion when 'SSO' is used in the generic sense. This commit aims to: - Rename `sso_` site settings. DiscourseConnect specific ones are prefixed `discourse_connect_`. Generic settings are prefixed `auth_` - Add (server-side-only) backwards compatibility for the old setting names, with deprecation notices - Copy `site_settings` database records to the new names - Rename relevant translation keys - Update relevant translations This commit does **not** aim to: - Rename any Ruby classes or methods. This might be done in a future commit - Change any URLs. This would break existing integrations - Make any changes to the protocol. This would break existing integrations - Change any functionality. Further normalization across DiscourseConnect and other auth methods will be done separately The risks are: - There is no backwards compatibility for site settings on the client-side. Accessing auth-related site settings in Javascript is fairly rare, and an error on the client side would not be security-critical. - If a plugin is monkey-patching parts of the auth process, changes to locale keys could cause broken error messages. This should also be unlikely. The old site setting names remain functional, so security-related overrides will remain working. A follow-up commit will be made with a post-deploy migration to delete the old `site_settings` rows.
2021-02-08 18:04:33 +08:00
SiteSetting.discourse_connect_url = "https://www.example.com/sso"
SiteSetting.enable_discourse_connect = true
end
it "only works for admins" do
get "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(403)
user.update(admin: true)
get "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
end
end
context "with missing token" do
it "returns the right response" do
get "/session/email-login"
expect(response.status).to eq(404)
end
end
context "with valid token" do
it "returns information" do
get "/session/email-login/#{email_token.token}.json"
expect(response.parsed_body["can_login"]).to eq(true)
expect(response.parsed_body["second_factor_required"]).to eq(nil)
# Does not log in the user
expect(session[:current_user_id]).to be_nil
end
it "fails when local logins via email is disabled" do
SiteSetting.enable_local_logins_via_email = false
get "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(403)
end
it "fails when local logins is disabled" do
SiteSetting.enable_local_logins = false
get "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(403)
end
context "when user has 2-factor logins" do
let!(:user_second_factor) { Fabricate(:user_second_factor_totp, user: user) }
let!(:user_second_factor_backup) { Fabricate(:user_second_factor_backup, user: user) }
it "includes that information in the response" do
get "/session/email-login/#{email_token.token}.json"
response_body_parsed = response.parsed_body
expect(response_body_parsed["can_login"]).to eq(true)
expect(response_body_parsed["second_factor_required"]).to eq(true)
expect(response_body_parsed["backup_codes_enabled"]).to eq(true)
end
end
context "when user has security key enabled" do
let!(:user_security_key) { Fabricate(:user_security_key, user: user) }
it "includes that information in the response" do
get "/session/email-login/#{email_token.token}.json"
response_body_parsed = response.parsed_body
expect(response_body_parsed["can_login"]).to eq(true)
expect(response_body_parsed["security_key_required"]).to eq(true)
expect(response_body_parsed["second_factor_required"]).to eq(nil)
expect(response_body_parsed["backup_codes_enabled"]).to eq(nil)
expect(response_body_parsed["allowed_credential_ids"]).to eq(
[user_security_key.credential_id],
)
secure_session = SecureSession.new(session["secure_session_id"])
expect(response_body_parsed["challenge"]).to eq(Webauthn.challenge(user, secure_session))
expect(Webauthn.rp_id(user, secure_session)).to eq(Discourse.current_hostname)
end
end
end
end
describe "#email_login" do
let(:email_token) do
Fabricate(:email_token, user: user, scope: EmailToken.scopes[:email_login])
end
before { SiteSetting.enable_local_logins_via_email = true }
context "when in staff writes only mode" do
use_redis_snapshotting
before { Discourse.enable_readonly_mode(Discourse::STAFF_WRITES_ONLY_MODE_KEY) }
it "allows admins to login" do
user.update!(admin: true)
post "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(200)
expect(session[:current_user_id]).to eq(user.id)
end
it "does not allow other users to login" do
post "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(503)
expect(session[:current_user_id]).to eq(nil)
end
end
context "when local logins via email disabled" do
before { SiteSetting.enable_local_logins_via_email = false }
it "only works for admins" do
post "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(403)
user.update(admin: true)
post "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(session[:current_user_id]).to eq(user.id)
end
end
context "with missing token" do
it "returns the right response" do
post "/session/email-login"
expect(response.status).to eq(404)
end
end
context "with invalid token" do
it "returns the right response" do
post "/session/email-login/adasdad.json"
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(
I18n.t("email_login.invalid_token", base_url: Discourse.base_url),
)
end
context "when token has expired" do
it "should return the right response" do
email_token.update!(created_at: 999.years.ago)
post "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(
I18n.t("email_login.invalid_token", base_url: Discourse.base_url),
)
end
end
end
context "with valid token" do
it "returns success" do
post "/session/email-login/#{email_token.token}.json"
expect(response.parsed_body["success"]).to eq("OK")
expect(session[:current_user_id]).to eq(user.id)
end
it "fails when local logins via email is disabled" do
SiteSetting.enable_local_logins_via_email = false
post "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(403)
expect(session[:current_user_id]).to eq(nil)
end
it "fails when local logins is disabled" do
SiteSetting.enable_local_logins = false
post "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(403)
expect(session[:current_user_id]).to eq(nil)
end
it "doesn't log in the user when not approved" do
SiteSetting.must_approve_users = true
post "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(I18n.t("login.not_approved"))
expect(session[:current_user_id]).to eq(nil)
end
context "when admin IP address is not valid" do
before do
Fabricate(
:screened_ip_address,
ip_address: "111.111.11.11",
action_type: ScreenedIpAddress.actions[:allow_admin],
)
SiteSetting.use_admin_ip_allowlist = true
user.update!(admin: true)
end
it "returns the right response" do
post "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(
I18n.t("login.admin_not_allowed_from_ip_address", username: user.username),
)
expect(session[:current_user_id]).to eq(nil)
end
end
context "when IP address is blocked" do
let(:permitted_ip_address) { "111.234.23.11" }
before do
Fabricate(
:screened_ip_address,
ip_address: permitted_ip_address,
action_type: ScreenedIpAddress.actions[:block],
)
end
it "returns the right response" do
ActionDispatch::Request.any_instance.stubs(:remote_ip).returns(permitted_ip_address)
post "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(
I18n.t("login.not_allowed_from_ip_address", username: user.username),
)
expect(session[:current_user_id]).to eq(nil)
end
end
context "when timezone param is provided" do
it "sets the user_option timezone for the user" do
post "/session/email-login/#{email_token.token}.json",
params: {
timezone: "Australia/Melbourne",
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(user.reload.user_option.timezone).to eq("Australia/Melbourne")
end
end
it "fails when user is suspended" do
user.update!(suspended_till: 2.days.from_now, suspended_at: Time.zone.now)
post "/session/email-login/#{email_token.token}.json"
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(
I18n.t("login.suspended", date: I18n.l(user.suspended_till, format: :date_only)),
)
expect(session[:current_user_id]).to eq(nil)
end
context "when user has 2-factor logins" do
2018-06-28 16:12:32 +08:00
let!(:user_second_factor) { Fabricate(:user_second_factor_totp, user: user) }
let!(:user_second_factor_backup) { Fabricate(:user_second_factor_backup, user: user) }
describe "errors on incorrect 2-factor" do
2018-06-28 16:12:32 +08:00
context "when using totp method" do
it "does not log in with incorrect two factor" do
post "/session/email-login/#{email_token.token}.json",
params: {
2018-06-28 16:12:32 +08:00
second_factor_token: "0000",
second_factor_method: UserSecondFactor.methods[:totp],
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(
I18n.t("login.invalid_second_factor_code"),
)
expect(session[:current_user_id]).to eq(nil)
2018-06-28 16:12:32 +08:00
end
end
context "when using backup code method" do
it "does not log in with incorrect backup code" do
post "/session/email-login/#{email_token.token}.json",
params: {
2018-06-28 16:12:32 +08:00
second_factor_token: "0000",
second_factor_method: UserSecondFactor.methods[:backup_codes],
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(
I18n.t("login.invalid_second_factor_code"),
)
expect(session[:current_user_id]).to eq(nil)
2018-06-28 16:12:32 +08:00
end
end
end
describe "allows successful 2-factor" do
2018-06-28 16:12:32 +08:00
context "when using totp method" do
it "logs in correctly" do
post "/session/email-login/#{email_token.token}.json",
params: {
2018-06-28 16:12:32 +08:00
second_factor_token: ROTP::TOTP.new(user_second_factor.data).now,
second_factor_method: UserSecondFactor.methods[:totp],
}
expect(response.parsed_body["success"]).to eq("OK")
expect(session[:current_user_id]).to eq(user.id)
2018-06-28 16:12:32 +08:00
end
end
context "when using backup code method" do
it "logs in correctly" do
post "/session/email-login/#{email_token.token}.json",
params: {
2018-06-28 16:12:32 +08:00
second_factor_token: "iAmValidBackupCode",
second_factor_method: UserSecondFactor.methods[:backup_codes],
}
expect(response.parsed_body["success"]).to eq("OK")
expect(session[:current_user_id]).to eq(user.id)
2018-06-28 16:12:32 +08:00
end
end
end
2020-01-15 18:27:12 +08:00
context "if the security_key_param is provided but only TOTP is enabled" do
it "does not log in the user" do
post "/session/email-login/#{email_token.token}.json",
params: {
second_factor_token: "foo",
second_factor_method: UserSecondFactor.methods[:totp],
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(I18n.t("login.invalid_second_factor_code"))
2020-01-15 18:27:12 +08:00
expect(session[:current_user_id]).to eq(nil)
end
end
end
context "when user has only security key enabled" do
let!(:user_security_key) do
Fabricate(
:user_security_key,
user: user,
credential_id: valid_security_key_data[:credential_id],
public_key: valid_security_key_data[:public_key],
)
end
before do
simulate_localhost_webauthn_challenge
# store challenge in secure session by visiting the email login page
get "/session/email-login/#{email_token.token}.json"
end
context "when the security key params are blank and a random second factor token is provided" do
it "shows an error message and denies login" do
post "/session/email-login/#{email_token.token}.json",
params: {
second_factor_token: "XXXXXXX",
second_factor_method: UserSecondFactor.methods[:totp],
}
expect(response.status).to eq(200)
expect(session[:current_user_id]).to eq(nil)
response_body = response.parsed_body
expect(response_body["error"]).to eq(I18n.t("login.not_enabled_second_factor_method"))
end
end
context "when the security key params are invalid" do
it "shows an error message and denies login" do
post "/session/email-login/#{email_token.token}.json",
params: {
2020-01-15 18:27:12 +08:00
second_factor_token: {
signature: "bad_sig",
clientData: "bad_clientData",
credentialId: "bad_credential_id",
authenticatorData: "bad_authenticator_data",
},
second_factor_method: UserSecondFactor.methods[:security_key],
}
expect(response.status).to eq(200)
expect(session[:current_user_id]).to eq(nil)
response_body = response.parsed_body
expect(response_body["failed"]).to eq("FAILED")
expect(response_body["error"]).to eq(I18n.t("webauthn.validation.not_found_error"))
end
end
context "when the security key params are valid" do
it "logs the user in" do
post "/session/email-login/#{email_token.token}.json",
params: {
login: user.username,
password: "myawesomepassword",
2020-01-15 18:27:12 +08:00
second_factor_token: valid_security_key_auth_post_data,
second_factor_method: UserSecondFactor.methods[:security_key],
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
user.reload
expect(session[:current_user_id]).to eq(user.id)
expect(user.user_auth_tokens.count).to eq(1)
end
end
end
2020-01-15 18:27:12 +08:00
context "when user has security key and totp enabled" do
2020-01-15 18:27:12 +08:00
let!(:user_security_key) do
Fabricate(
:user_security_key,
user: user,
credential_id: valid_security_key_data[:credential_id],
public_key: valid_security_key_data[:public_key],
)
end
let!(:user_second_factor) { Fabricate(:user_second_factor_totp, user: user) }
it "doesnt allow logging in if the 2fa params are garbled" do
post "/session/email-login/#{email_token.token}.json",
params: {
second_factor_method: UserSecondFactor.methods[:totp],
second_factor_token: "blah",
}
expect(response.status).to eq(200)
expect(session[:current_user_id]).to eq(nil)
response_body = response.parsed_body
2020-01-15 18:27:12 +08:00
expect(response_body["error"]).to eq(I18n.t("login.invalid_second_factor_code"))
end
it "doesnt allow login if both of the 2fa params are blank" do
post "/session/email-login/#{email_token.token}.json",
params: {
second_factor_method: UserSecondFactor.methods[:totp],
second_factor_token: "",
}
expect(response.status).to eq(200)
expect(session[:current_user_id]).to eq(nil)
response_body = response.parsed_body
2020-01-15 18:27:12 +08:00
expect(response_body["error"]).to eq(I18n.t("login.invalid_second_factor_code"))
end
end
end
end
describe "logoff support" do
it "can log off users cleanly" do
user = Fabricate(:user)
sign_in(user)
UserAuthToken.destroy_all
# we need a route that will call current user
post "/drafts.json", params: {}
expect(response.headers["Discourse-Logged-Out"]).to eq("1")
end
end
describe "#become" do
let!(:user) { Fabricate(:user) }
it "does not work when in production mode" do
Rails.env.stubs(:production?).returns(true)
get "/session/#{user.username}/become.json"
expect(response.status).to eq(403)
expect(response.parsed_body["error_type"]).to eq("invalid_access")
expect(session[:current_user_id]).to be_blank
end
it "works in development mode" do
Rails.env.stubs(:development?).returns(true)
get "/session/#{user.username}/become.json"
expect(response).to be_redirect
expect(session[:current_user_id]).to eq(user.id)
end
end
describe "#sso" do
before do
SiteSetting.discourse_connect_url = "http://example.com/discourse_sso"
SiteSetting.enable_discourse_connect = true
SiteSetting.discourse_connect_secret = "shjkfdhsfkjh"
end
it "redirects correctly" do
get "/session/sso"
expect(response.status).to eq(302)
expect(response.location).to start_with(SiteSetting.discourse_connect_url)
end
end
describe "#sso_login" do
before do
@sso_url = "http://example.com/discourse_sso"
@sso_secret = "shjkfdhsfkjh"
FEATURE: Rename 'Discourse SSO' to DiscourseConnect (#11978) The 'Discourse SSO' protocol is being rebranded to DiscourseConnect. This should help to reduce confusion when 'SSO' is used in the generic sense. This commit aims to: - Rename `sso_` site settings. DiscourseConnect specific ones are prefixed `discourse_connect_`. Generic settings are prefixed `auth_` - Add (server-side-only) backwards compatibility for the old setting names, with deprecation notices - Copy `site_settings` database records to the new names - Rename relevant translation keys - Update relevant translations This commit does **not** aim to: - Rename any Ruby classes or methods. This might be done in a future commit - Change any URLs. This would break existing integrations - Make any changes to the protocol. This would break existing integrations - Change any functionality. Further normalization across DiscourseConnect and other auth methods will be done separately The risks are: - There is no backwards compatibility for site settings on the client-side. Accessing auth-related site settings in Javascript is fairly rare, and an error on the client side would not be security-critical. - If a plugin is monkey-patching parts of the auth process, changes to locale keys could cause broken error messages. This should also be unlikely. The old site setting names remain functional, so security-related overrides will remain working. A follow-up commit will be made with a post-deploy migration to delete the old `site_settings` rows.
2021-02-08 18:04:33 +08:00
SiteSetting.discourse_connect_url = @sso_url
SiteSetting.enable_discourse_connect = true
SiteSetting.discourse_connect_secret = @sso_secret
Fabricate(:admin)
end
let(:headers) { { host: Discourse.current_hostname } }
def get_sso(return_path)
nonce = SecureRandom.hex
dso = DiscourseConnect.new(secure_session: read_secure_session)
dso.nonce = nonce
dso.register_nonce(return_path)
sso = DiscourseConnectBase.new
sso.nonce = nonce
sso.sso_secret = @sso_secret
sso
end
context "when in staff writes only mode" do
use_redis_snapshotting
before { Discourse.enable_readonly_mode(Discourse::STAFF_WRITES_ONLY_MODE_KEY) }
it "allows staff to login" do
sso = get_sso("/a/")
sso.external_id = "666"
sso.email = "bob@bob.com"
sso.name = "Bob Bobson"
sso.username = "bob"
sso.admin = true
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user).not_to eq(nil)
end
it 'doesn\'t allow non-staff to login' do
sso = get_sso("/a/")
sso.external_id = "666"
sso.email = "bob@bob.com"
sso.name = "Bob Bobson"
sso.username = "bob"
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user).to eq(nil)
end
end
it "does not create superfluous auth tokens when already logged in" do
user = Fabricate(:user)
sign_in(user)
sso = get_sso("/")
sso.email = user.email
sso.external_id = "abc"
sso.username = "sam"
expect do
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user.id).to eq(user.id)
end.not_to change { UserAuthToken.count }
end
it "will never redirect back to /session/sso path" do
sso = get_sso("/session/sso?bla=1")
sso.email = user.email
sso.external_id = "abc"
sso.username = "sam"
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(response).to redirect_to("/")
sso = get_sso("http://#{Discourse.current_hostname}/session/sso?bla=1")
sso.email = user.email
sso.external_id = "abc"
sso.username = "sam"
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(response).to redirect_to("/")
end
it "can handle invalid sso external ids due to blank" do
sso = get_sso("/")
sso.email = "test@test.com"
sso.external_id = " "
sso.username = "sam"
logger =
track_log_messages do
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
end
expect(logger.warnings.length).to eq(0)
expect(logger.errors.length).to eq(0)
expect(logger.fatals.length).to eq(0)
expect(response.status).to eq(500)
FEATURE: Rename 'Discourse SSO' to DiscourseConnect (#11978) The 'Discourse SSO' protocol is being rebranded to DiscourseConnect. This should help to reduce confusion when 'SSO' is used in the generic sense. This commit aims to: - Rename `sso_` site settings. DiscourseConnect specific ones are prefixed `discourse_connect_`. Generic settings are prefixed `auth_` - Add (server-side-only) backwards compatibility for the old setting names, with deprecation notices - Copy `site_settings` database records to the new names - Rename relevant translation keys - Update relevant translations This commit does **not** aim to: - Rename any Ruby classes or methods. This might be done in a future commit - Change any URLs. This would break existing integrations - Make any changes to the protocol. This would break existing integrations - Change any functionality. Further normalization across DiscourseConnect and other auth methods will be done separately The risks are: - There is no backwards compatibility for site settings on the client-side. Accessing auth-related site settings in Javascript is fairly rare, and an error on the client side would not be security-critical. - If a plugin is monkey-patching parts of the auth process, changes to locale keys could cause broken error messages. This should also be unlikely. The old site setting names remain functional, so security-related overrides will remain working. A follow-up commit will be made with a post-deploy migration to delete the old `site_settings` rows.
2021-02-08 18:04:33 +08:00
expect(response.body).to include(I18n.t("discourse_connect.blank_id_error"))
end
it "can handle invalid sso email validation errors" do
SiteSetting.blocked_email_domains = "test.com"
sso = get_sso("/")
sso.email = "test@test.com"
sso.external_id = "123"
sso.username = "sam"
logger =
track_log_messages do
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
end
expect(logger.warnings.length).to eq(0)
expect(logger.errors.length).to eq(0)
expect(logger.fatals.length).to eq(0)
expect(response.status).to eq(500)
FEATURE: Rename 'Discourse SSO' to DiscourseConnect (#11978) The 'Discourse SSO' protocol is being rebranded to DiscourseConnect. This should help to reduce confusion when 'SSO' is used in the generic sense. This commit aims to: - Rename `sso_` site settings. DiscourseConnect specific ones are prefixed `discourse_connect_`. Generic settings are prefixed `auth_` - Add (server-side-only) backwards compatibility for the old setting names, with deprecation notices - Copy `site_settings` database records to the new names - Rename relevant translation keys - Update relevant translations This commit does **not** aim to: - Rename any Ruby classes or methods. This might be done in a future commit - Change any URLs. This would break existing integrations - Make any changes to the protocol. This would break existing integrations - Change any functionality. Further normalization across DiscourseConnect and other auth methods will be done separately The risks are: - There is no backwards compatibility for site settings on the client-side. Accessing auth-related site settings in Javascript is fairly rare, and an error on the client side would not be security-critical. - If a plugin is monkey-patching parts of the auth process, changes to locale keys could cause broken error messages. This should also be unlikely. The old site setting names remain functional, so security-related overrides will remain working. A follow-up commit will be made with a post-deploy migration to delete the old `site_settings` rows.
2021-02-08 18:04:33 +08:00
expect(response.body).to include(
I18n.t("discourse_connect.email_error", email: ERB::Util.html_escape("test@test.com")),
)
end
it "can handle invalid sso external ids due to banned word" do
sso = get_sso("/")
sso.email = "test@test.com"
sso.external_id = "nil"
sso.username = "sam"
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(response.status).to eq(500)
end
it "can take over an account" do
user = Fabricate(:user, email: "bill@bill.com")
sso = get_sso("/")
sso.email = user.email
sso.external_id = "abc"
sso.username = "sam"
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(response).to redirect_to("/")
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user.email).to eq(user.email)
expect(logged_on_user.single_sign_on_record.external_id).to eq("abc")
expect(logged_on_user.single_sign_on_record.external_username).to eq("sam")
# we are updating the email ... ensure auto group membership works
sign_out
SiteSetting.email_editable = false
FEATURE: Rename 'Discourse SSO' to DiscourseConnect (#11978) The 'Discourse SSO' protocol is being rebranded to DiscourseConnect. This should help to reduce confusion when 'SSO' is used in the generic sense. This commit aims to: - Rename `sso_` site settings. DiscourseConnect specific ones are prefixed `discourse_connect_`. Generic settings are prefixed `auth_` - Add (server-side-only) backwards compatibility for the old setting names, with deprecation notices - Copy `site_settings` database records to the new names - Rename relevant translation keys - Update relevant translations This commit does **not** aim to: - Rename any Ruby classes or methods. This might be done in a future commit - Change any URLs. This would break existing integrations - Make any changes to the protocol. This would break existing integrations - Change any functionality. Further normalization across DiscourseConnect and other auth methods will be done separately The risks are: - There is no backwards compatibility for site settings on the client-side. Accessing auth-related site settings in Javascript is fairly rare, and an error on the client side would not be security-critical. - If a plugin is monkey-patching parts of the auth process, changes to locale keys could cause broken error messages. This should also be unlikely. The old site setting names remain functional, so security-related overrides will remain working. A follow-up commit will be made with a post-deploy migration to delete the old `site_settings` rows.
2021-02-08 18:04:33 +08:00
SiteSetting.auth_overrides_email = true
group = Fabricate(:group, name: :bob, automatic_membership_email_domains: "jane.com")
sso = get_sso("/")
sso.email = "hello@jane.com"
sso.external_id = "abc"
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user.email).to eq("hello@jane.com")
expect(group.users.count).to eq(1)
end
def sso_for_ip_specs
sso = get_sso("/a/")
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
sso.external_id = "666"
sso.email = "bob@bob.com"
sso.name = "Sam Saffron"
sso.username = "sam"
sso
end
it "respects IP restrictions on create" do
ScreenedIpAddress.all.destroy_all
get "/"
_screened_ip =
Fabricate(
:screened_ip_address,
ip_address: request.remote_ip,
action_type: ScreenedIpAddress.actions[:block],
)
sso = sso_for_ip_specs
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user).to eq(nil)
end
it "respects IP restrictions on login" do
ScreenedIpAddress.all.destroy_all
get "/"
sso = sso_for_ip_specs
DiscourseConnect.parse(
sso.payload,
secure_session: read_secure_session,
).lookup_or_create_user(request.remote_ip)
sso = sso_for_ip_specs
_screened_ip =
Fabricate(
:screened_ip_address,
ip_address: request.remote_ip,
action_type: ScreenedIpAddress.actions[:block],
)
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user).to be_blank
end
it "respects email restrictions" do
sso = get_sso("/a/")
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
sso.external_id = "666"
sso.email = "bob@bob.com"
sso.name = "Sam Saffron"
sso.username = "sam"
ScreenedEmail.block("bob@bob.com")
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user).to eq(nil)
end
it "allows you to create an admin account" do
sso = get_sso("/a/")
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
sso.external_id = "666"
sso.email = "bob@bob.com"
sso.name = "Sam Saffron"
sso.username = "sam"
sso.custom_fields["shop_url"] = "http://my_shop.com"
sso.custom_fields["shop_name"] = "Sam"
sso.admin = true
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user.admin).to eq(true)
end
it "does not redirect offsite" do
sso = get_sso("#{Discourse.base_url}//site.com/xyz")
sso.external_id = "666"
sso.email = "bob@bob.com"
sso.name = "Sam Saffron"
sso.username = "sam"
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(response).to redirect_to("#{Discourse.base_url}//site.com/xyz")
end
it "redirects to a non-relative url" do
sso = get_sso("#{Discourse.base_url}/b/")
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
sso.external_id = "666"
sso.email = "bob@bob.com"
sso.name = "Sam Saffron"
sso.username = "sam"
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(response).to redirect_to("/b/")
end
it "redirects to random url if it is allowed" do
FEATURE: Rename 'Discourse SSO' to DiscourseConnect (#11978) The 'Discourse SSO' protocol is being rebranded to DiscourseConnect. This should help to reduce confusion when 'SSO' is used in the generic sense. This commit aims to: - Rename `sso_` site settings. DiscourseConnect specific ones are prefixed `discourse_connect_`. Generic settings are prefixed `auth_` - Add (server-side-only) backwards compatibility for the old setting names, with deprecation notices - Copy `site_settings` database records to the new names - Rename relevant translation keys - Update relevant translations This commit does **not** aim to: - Rename any Ruby classes or methods. This might be done in a future commit - Change any URLs. This would break existing integrations - Make any changes to the protocol. This would break existing integrations - Change any functionality. Further normalization across DiscourseConnect and other auth methods will be done separately The risks are: - There is no backwards compatibility for site settings on the client-side. Accessing auth-related site settings in Javascript is fairly rare, and an error on the client side would not be security-critical. - If a plugin is monkey-patching parts of the auth process, changes to locale keys could cause broken error messages. This should also be unlikely. The old site setting names remain functional, so security-related overrides will remain working. A follow-up commit will be made with a post-deploy migration to delete the old `site_settings` rows.
2021-02-08 18:04:33 +08:00
SiteSetting.discourse_connect_allows_all_return_paths = true
sso = get_sso("https://gusundtrout.com")
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
sso.external_id = "666"
sso.email = "bob@bob.com"
sso.name = "Sam Saffron"
sso.username = "sam"
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(response).to redirect_to("https://gusundtrout.com")
end
it "redirects to root if the host of the return_path is different" do
sso = get_sso("//eviltrout.com")
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
sso.external_id = "666"
sso.email = "bob@bob.com"
sso.name = "Sam Saffron"
sso.username = "sam"
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(response).to redirect_to("/")
end
it "redirects to root if the host of the return_path is different" do
sso = get_sso("http://eviltrout.com")
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
sso.external_id = "666"
sso.email = "bob@bob.com"
sso.name = "Sam Saffron"
sso.username = "sam"
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(response).to redirect_to("/")
end
it "creates a user but ignores auto_approve_email_domains site setting when must_approve_users site setting is not enabled" do
SiteSetting.auto_approve_email_domains = "discourse.com"
sso = get_sso("/a/")
sso.external_id = "666"
sso.email = "sam@discourse.com"
sso.name = "Sam Saffron"
sso.username = "sam"
events =
DiscourseEvent.track_events do
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(response).to redirect_to("/a/")
end
expect(events.map { |event| event[:event_name] }).to include(
:user_logged_in,
:user_first_logged_in,
)
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
# ensure nothing is transient
logged_on_user = User.find(logged_on_user.id)
expect(logged_on_user.admin).to eq(false)
expect(logged_on_user.email).to eq("sam@discourse.com")
expect(logged_on_user.name).to eq("Sam Saffron")
expect(logged_on_user.username).to eq("sam")
expect(logged_on_user.approved).to eq(false)
expect(logged_on_user.active).to eq(true)
expect(logged_on_user.single_sign_on_record.external_id).to eq("666")
expect(logged_on_user.single_sign_on_record.external_username).to eq("sam")
end
context "when must_approve_users site setting has been enabled" do
before { SiteSetting.must_approve_users = true }
it "creates a user but does not approve when user's email domain does not match a domain in auto_approve_email_domains site settings" do
SiteSetting.auto_approve_email_domains = "discourse.com"
sso = get_sso("/a/")
sso.external_id = "666"
sso.email = "sam@discourse.org"
sso.name = "Sam Saffron"
sso.username = "sam"
expect do
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(response.status).to eq(403)
expect(response.parsed_body).to include(I18n.t("discourse_connect.account_not_approved"))
end.to change { User.count }.by(1)
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user).to eq(nil)
user = User.last
expect(user.admin).to eq(false)
expect(user.email).to eq("sam@discourse.org")
expect(user.name).to eq("Sam Saffron")
expect(user.username).to eq("sam")
expect(user.approved).to eq(false)
expect(user.active).to eq(true)
expect(user.single_sign_on_record.external_id).to eq("666")
expect(user.single_sign_on_record.external_username).to eq("sam")
end
it "creates and approves a user when user's email domain matches a domain in auto_approve_email_domains site settings" do
SiteSetting.auto_approve_email_domains = "discourse.com"
sso = get_sso("/a/")
sso.external_id = "666"
sso.email = "sam@discourse.com"
sso.name = "Sam Saffron"
sso.username = "sam"
events =
DiscourseEvent.track_events do
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(response).to redirect_to("/a/")
end
expect(events.map { |event| event[:event_name] }).to include(
:user_logged_in,
:user_first_logged_in,
)
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
# ensure nothing is transient
logged_on_user = User.find(logged_on_user.id)
expect(logged_on_user.admin).to eq(false)
expect(logged_on_user.email).to eq("sam@discourse.com")
expect(logged_on_user.name).to eq("Sam Saffron")
expect(logged_on_user.username).to eq("sam")
expect(logged_on_user.approved).to eq(true)
expect(logged_on_user.active).to eq(true)
expect(logged_on_user.single_sign_on_record.external_id).to eq("666")
expect(logged_on_user.single_sign_on_record.external_username).to eq("sam")
end
end
it "allows you to create an account" do
group = Fabricate(:group, name: :bob, automatic_membership_email_domains: "bob.com")
sso = get_sso("/a/")
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
sso.external_id = "666"
sso.email = "bob@bob.com"
sso.name = "Sam Saffron"
sso.username = "sam"
sso.custom_fields["shop_url"] = "http://my_shop.com"
sso.custom_fields["shop_name"] = "Sam"
events =
DiscourseEvent.track_events do
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
end
expect(events.map { |event| event[:event_name] }).to include(
:user_logged_in,
:user_first_logged_in,
)
expect(response).to redirect_to("/a/")
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(group.users.where(id: logged_on_user.id).count).to eq(1)
# ensure nothing is transient
logged_on_user = User.find(logged_on_user.id)
expect(logged_on_user.admin).to eq(false)
expect(logged_on_user.email).to eq("bob@bob.com")
expect(logged_on_user.name).to eq("Sam Saffron")
expect(logged_on_user.username).to eq("sam")
expect(logged_on_user.single_sign_on_record.external_id).to eq("666")
expect(logged_on_user.single_sign_on_record.external_username).to eq("sam")
expect(logged_on_user.active).to eq(true)
expect(logged_on_user.custom_fields["shop_url"]).to eq("http://my_shop.com")
expect(logged_on_user.custom_fields["shop_name"]).to eq("Sam")
expect(logged_on_user.custom_fields["bla"]).to eq(nil)
end
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
context "when an invitation is used" do
let(:invite) { Fabricate(:invite, email: invite_email, invited_by: Fabricate(:admin)) }
let(:invite_email) { nil }
def login_with_sso_and_invite(invite_key = invite.invite_key)
write_secure_session("invite-key", invite_key)
sso = get_sso("/")
sso.external_id = "666"
sso.email = "bob@bob.com"
sso.name = "Sam Saffron"
sso.username = "sam"
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
end
it "errors if the invite key is invalid" do
login_with_sso_and_invite("wrong")
expect(response.status).to eq(400)
expect(response.body).to include(I18n.t("invite.not_found", base_url: Discourse.base_url))
expect(invite.reload.redeemed?).to eq(false)
expect(User.find_by_email("bob@bob.com")).to eq(nil)
end
it "errors if the invite has expired" do
invite.update!(expires_at: 3.days.ago)
login_with_sso_and_invite
expect(response.status).to eq(400)
expect(response.body).to include(I18n.t("invite.expired", base_url: Discourse.base_url))
expect(invite.reload.redeemed?).to eq(false)
expect(User.find_by_email("bob@bob.com")).to eq(nil)
end
it "errors if the invite has been redeemed already" do
invite.update!(max_redemptions_allowed: 1, redemption_count: 1)
login_with_sso_and_invite
expect(response.status).to eq(400)
expect(response.body).to include(
I18n.t(
"invite.not_found_template",
site_name: SiteSetting.title,
base_url: Discourse.base_url,
),
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
)
expect(invite.reload.redeemed?).to eq(true)
expect(User.find_by_email("bob@bob.com")).to eq(nil)
end
it "errors if the invite is for a specific email and that email does not match the sso email" do
invite.update!(email: "someotheremail@dave.com")
login_with_sso_and_invite
expect(response.status).to eq(400)
expect(response.body).to include(
I18n.t("invite.not_matching_email", base_url: Discourse.base_url),
)
expect(invite.reload.redeemed?).to eq(false)
expect(User.find_by_email("bob@bob.com")).to eq(nil)
end
it "allows you to create an account and redeems the invite successfully, clearing the invite-key session" do
login_with_sso_and_invite
expect(response.status).to eq(302)
expect(response).to redirect_to("/")
expect(invite.reload.redeemed?).to eq(true)
user = User.find_by_email("bob@bob.com")
expect(user.active).to eq(true)
expect(session[:current_user_id]).to eq(user.id)
expect(read_secure_session["invite-key"]).to eq(nil)
end
it "creates the user account and redeems the invite but does not approve the user if must_approve_users is enabled" do
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
SiteSetting.must_approve_users = true
login_with_sso_and_invite
expect(response.status).to eq(403)
expect(response.parsed_body).to include(I18n.t("discourse_connect.account_not_approved"))
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
expect(invite.reload.redeemed?).to eq(true)
user = User.find_by_email("bob@bob.com")
expect(user.active).to eq(true)
expect(user.approved).to eq(false)
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
end
it "redirects to the topic associated to the invite" do
topic_invite = TopicInvite.create!(invite: invite, topic: Fabricate(:topic))
login_with_sso_and_invite
expect(response.status).to eq(302)
expect(response).to redirect_to(topic_invite.topic.relative_url)
end
it "adds the user to the appropriate invite groups" do
invited_group = InvitedGroup.create!(invite: invite, group: Fabricate(:group))
login_with_sso_and_invite
expect(invite.reload.redeemed?).to eq(true)
user = User.find_by_email("bob@bob.com")
expect(GroupUser.exists?(user: user, group: invited_group.group)).to eq(true)
end
end
context "when sso emails are not trusted" do
context "if you have not activated your account" do
it "does not log you in" do
sso = get_sso("/a/")
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
sso.external_id = "666"
sso.email = "bob@bob.com"
sso.name = "Sam Saffron"
sso.username = "sam"
sso.require_activation = true
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user).to eq(nil)
end
it "sends an activation email" do
sso = get_sso("/a/")
FEATURE: Allow using invites when DiscourseConnect SSO is enabled (#12419) This PR allows invitations to be used when the DiscourseConnect SSO is enabled for a site (`enable_discourse_connect`) and local logins are disabled. Previously invites could not be accepted with SSO enabled simply because we did not have the code paths to handle that logic. The invitation methods that are supported include: * Inviting people to groups via email address * Inviting people to topics via email address * Using invitation links generated by the Invite Users UI in the /my/invited/pending route The flow works like this: 1. User visits an invite URL 2. The normal invitation validations (redemptions/expiry) happen at that point 3. We store the invite key in a secure session 4. The user clicks "Accept Invitation and Continue" (see below) 5. The user is redirected to /session/sso then to the SSO provider URL then back to /session/sso_login 6. We retrieve the invite based on the invite key in secure session. We revalidate the invitation. We show an error to the user if it is not valid. An additional check here for invites with an email specified is to check the SSO email matches the invite email 7. If the invite is OK we create the user via the normal SSO methods 8. We redeem the invite and activate the user. We clear the invite key in secure session. 9. If the invite had a topic we redirect the user there, otherwise we redirect to / Note that we decided for SSO-based invites the `must_approve_users` site setting is ignored, because the invite is a form of pre-approval, and because regular non-staff users cannot send out email invites or generally invite to the forum in this case. Also deletes some group invite checks as per https://github.com/discourse/discourse/pull/12353
2021-03-19 08:20:10 +08:00
sso.external_id = "666"
sso.email = "bob@bob.com"
sso.name = "Sam Saffron"
sso.username = "sam"
sso.require_activation = true
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(Jobs::CriticalUserEmail.jobs.size).to eq(1)
end
end
context "if you have activated your account" do
it "allows you to log in" do
sso = get_sso("/hello/world")
sso.external_id = "997"
sso.sso_url = "http://somewhere.over.com/sso_login"
sso.require_activation = true
user = Fabricate(:user)
user.create_single_sign_on_record(external_id: "997", last_payload: "")
user.stubs(:active?).returns(true)
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(user.id).to eq(logged_on_user.id)
end
end
end
it "allows login to existing account with valid nonce" do
sso = get_sso("/hello/world")
sso.external_id = "997"
sso.sso_url = "http://somewhere.over.com/sso_login"
user = Fabricate(:user)
user.create_single_sign_on_record(external_id: "997", last_payload: "")
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
user.single_sign_on_record.reload
expect(user.single_sign_on_record.last_payload).to eq(sso.unsigned_payload)
expect(response).to redirect_to("/hello/world")
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(user.id).to eq(logged_on_user.id)
# nonce is bad now
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(response.status).to eq(419)
end
it "associates the nonce with the current session" do
sso = get_sso("/hello/world")
sso.external_id = "997"
sso.sso_url = "http://somewhere.over.com/sso_login"
user = Fabricate(:user)
user.create_single_sign_on_record(external_id: "997", last_payload: "")
# Establish a fresh session
cookies.to_hash.keys.each { |k| cookies.delete(k) }
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
expect(response.status).to eq(419)
end
context "when sso provider is enabled" do
before do
FEATURE: Rename 'Discourse SSO' to DiscourseConnect (#11978) The 'Discourse SSO' protocol is being rebranded to DiscourseConnect. This should help to reduce confusion when 'SSO' is used in the generic sense. This commit aims to: - Rename `sso_` site settings. DiscourseConnect specific ones are prefixed `discourse_connect_`. Generic settings are prefixed `auth_` - Add (server-side-only) backwards compatibility for the old setting names, with deprecation notices - Copy `site_settings` database records to the new names - Rename relevant translation keys - Update relevant translations This commit does **not** aim to: - Rename any Ruby classes or methods. This might be done in a future commit - Change any URLs. This would break existing integrations - Make any changes to the protocol. This would break existing integrations - Change any functionality. Further normalization across DiscourseConnect and other auth methods will be done separately The risks are: - There is no backwards compatibility for site settings on the client-side. Accessing auth-related site settings in Javascript is fairly rare, and an error on the client side would not be security-critical. - If a plugin is monkey-patching parts of the auth process, changes to locale keys could cause broken error messages. This should also be unlikely. The old site setting names remain functional, so security-related overrides will remain working. A follow-up commit will be made with a post-deploy migration to delete the old `site_settings` rows.
2021-02-08 18:04:33 +08:00
SiteSetting.enable_discourse_connect_provider = true
SiteSetting.discourse_connect_provider_secrets = %w[
*|secret,forAll
*.rainbow|wrongSecretForOverRainbow
www.random.site|secretForRandomSite
somewhere.over.rainbow|secretForOverRainbow
].join("\n")
end
it "doesn't break" do
sso = get_sso("/hello/world")
sso.external_id = "997"
sso.sso_url = "http://somewhere.over.com/sso_login"
sso.return_sso_url = "http://someurl.com"
user = Fabricate(:user)
user.create_single_sign_on_record(external_id: "997", last_payload: "")
get "/session/sso_login", params: Rack::Utils.parse_query(sso.payload), headers: headers
user.single_sign_on_record.reload
expect(user.single_sign_on_record.last_payload).to eq(sso.unsigned_payload)
expect(response).to redirect_to("/hello/world")
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(user.id).to eq(logged_on_user.id)
end
end
it "returns the correct error code for invalid signature" do
sso = get_sso("/hello/world")
sso.external_id = "997"
sso.sso_url = "http://somewhere.over.com/sso_login"
correct_params = Rack::Utils.parse_query(sso.payload)
get "/session/sso_login",
params: correct_params.merge(sig: "thisisnotthesigyouarelookingfor"),
headers: headers
expect(response.status).to eq(422)
expect(response.body).not_to include(correct_params["sig"]) # Check we didn't send the real sig back to the client
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user).to eq(nil)
correct_params = Rack::Utils.parse_query(sso.payload)
get "/session/sso_login",
params: correct_params.merge(sig: "thisisasignaturewith@special!characters"),
headers: headers
expect(response.status).to eq(422)
expect(response.body).not_to include(correct_params["sig"]) # Check we didn't send the real sig back to the client
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user).to eq(nil)
end
describe "local attribute override from SSO payload" do
before do
SiteSetting.email_editable = false
FEATURE: Rename 'Discourse SSO' to DiscourseConnect (#11978) The 'Discourse SSO' protocol is being rebranded to DiscourseConnect. This should help to reduce confusion when 'SSO' is used in the generic sense. This commit aims to: - Rename `sso_` site settings. DiscourseConnect specific ones are prefixed `discourse_connect_`. Generic settings are prefixed `auth_` - Add (server-side-only) backwards compatibility for the old setting names, with deprecation notices - Copy `site_settings` database records to the new names - Rename relevant translation keys - Update relevant translations This commit does **not** aim to: - Rename any Ruby classes or methods. This might be done in a future commit - Change any URLs. This would break existing integrations - Make any changes to the protocol. This would break existing integrations - Change any functionality. Further normalization across DiscourseConnect and other auth methods will be done separately The risks are: - There is no backwards compatibility for site settings on the client-side. Accessing auth-related site settings in Javascript is fairly rare, and an error on the client side would not be security-critical. - If a plugin is monkey-patching parts of the auth process, changes to locale keys could cause broken error messages. This should also be unlikely. The old site setting names remain functional, so security-related overrides will remain working. A follow-up commit will be made with a post-deploy migration to delete the old `site_settings` rows.
2021-02-08 18:04:33 +08:00
SiteSetting.auth_overrides_email = true
SiteSetting.auth_overrides_username = true
SiteSetting.auth_overrides_name = true
@user = Fabricate(:user)
@sso = get_sso("/hello/world")
@sso.external_id = "997"
@reversed_username = @user.username.reverse
@sso.username = @reversed_username
@sso.email = "#{@reversed_username}@garbage.org"
@reversed_name = @user.name.reverse
@sso.name = @reversed_name
@suggested_username = UserNameSuggester.suggest(@sso.username || @sso.name || @sso.email)
@suggested_name = User.suggest_name(@sso.name || @sso.username || @sso.email)
@user.create_single_sign_on_record(external_id: "997", last_payload: "")
end
it "stores the external attributes" do
get "/session/sso_login", params: Rack::Utils.parse_query(@sso.payload), headers: headers
@user.single_sign_on_record.reload
expect(@user.single_sign_on_record.external_username).to eq(@sso.username)
expect(@user.single_sign_on_record.external_email).to eq(@sso.email)
expect(@user.single_sign_on_record.external_name).to eq(@sso.name)
end
it "overrides attributes" do
get "/session/sso_login", params: Rack::Utils.parse_query(@sso.payload), headers: headers
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user.username).to eq(@suggested_username)
expect(logged_on_user.email).to eq("#{@reversed_username}@garbage.org")
expect(logged_on_user.name).to eq(@sso.name)
end
it "does not change matching attributes for an existing account" do
@sso.username = @user.username
@sso.name = @user.name
@sso.email = @user.email
get "/session/sso_login", params: Rack::Utils.parse_query(@sso.payload), headers: headers
logged_on_user = Discourse.current_user_provider.new(request.env).current_user
expect(logged_on_user.username).to eq(@user.username)
expect(logged_on_user.name).to eq(@user.name)
expect(logged_on_user.email).to eq(@user.email)
end
end
context "when in readonly mode" do
use_redis_snapshotting
before { Discourse.enable_readonly_mode }
it "disallows requests" do
get "/session/sso_login"
expect(response.status).to eq(503)
end
end
end
describe "#sso_provider" do
let(:headers) { { host: Discourse.current_hostname } }
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
let(:logo_fixture) { "http://#{Discourse.current_hostname}/uploads/logo.png" }
fab!(:user) { Fabricate(:user, password: "myfrogs123ADMIN", active: true, admin: true) }
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
before do
stub_request(:any, %r{#{Discourse.current_hostname}/uploads}).to_return(
status: 200,
body: lambda { |request| file_from_fixtures("logo.png") },
)
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
SiteSetting.enable_discourse_connect_provider = true
SiteSetting.enable_discourse_connect = false
SiteSetting.enable_local_logins = true
SiteSetting.discourse_connect_provider_secrets = %w[
*|secret,forAll
*.rainbow|wrongSecretForOverRainbow
www.random.site|secretForRandomSite
somewhere.over.rainbow|oldSecretForOverRainbow
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
somewhere.over.rainbow|secretForOverRainbow
somewhere.over.rainbow|newSecretForOverRainbow
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
].join("\n")
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
@sso = DiscourseConnectProvider.new
@sso.nonce = "mynonce"
@sso.return_sso_url = "http://somewhere.over.rainbow/sso"
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
@user = user
group = Fabricate(:group)
group.add(@user)
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
@user.create_user_avatar!
UserAvatar.import_url_for_user(logo_fixture, @user)
UserProfile.import_url_for_user(logo_fixture, @user, is_card_background: false)
UserProfile.import_url_for_user(logo_fixture, @user, is_card_background: true)
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
@user.reload
@user.user_avatar.reload
@user.user_profile.reload
EmailToken.update_all(confirmed: true)
end
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
describe "can act as an SSO provider" do
it "successfully logs in and redirects user to return_sso_url when the user is not logged in" do
get "/session/sso_provider",
params: Rack::Utils.parse_query(@sso.payload("secretForOverRainbow"))
expect(response).to redirect_to("/login")
post "/session.json",
params: {
login: @user.username,
password: "myfrogs123ADMIN",
},
xhr: true,
headers: headers
location = response.cookies["sso_destination_url"]
# javascript code will handle redirection of user to return_sso_url
expect(location).to match(%r{^http://somewhere.over.rainbow/sso})
payload = location.split("?")[1]
sso2 = DiscourseConnectProvider.parse(payload)
expect(sso2.email).to eq(@user.email)
expect(sso2.name).to eq(@user.name)
expect(sso2.username).to eq(@user.username)
expect(sso2.external_id).to eq(@user.id.to_s)
expect(sso2.admin).to eq(true)
expect(sso2.moderator).to eq(false)
expect(sso2.groups).to eq(@user.groups.pluck(:name).join(","))
expect(sso2.avatar_url.blank?).to_not eq(true)
expect(sso2.profile_background_url.blank?).to_not eq(true)
expect(sso2.card_background_url.blank?).to_not eq(true)
expect(sso2.avatar_url).to start_with(Discourse.base_url)
expect(sso2.profile_background_url).to start_with(Discourse.base_url)
expect(sso2.card_background_url).to start_with(Discourse.base_url)
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
expect(sso2.confirmed_2fa).to eq(nil)
expect(sso2.no_2fa_methods).to eq(nil)
end
it "correctly logs in for secondary domain secrets" do
sign_in @user
get "/session/sso_provider",
params: Rack::Utils.parse_query(@sso.payload("newSecretForOverRainbow"))
expect(response.status).to eq(302)
redirect_uri = URI.parse(response.location)
expect(redirect_uri.host).to eq("somewhere.over.rainbow")
redirect_query = CGI.parse(redirect_uri.query)
expected_sig =
DiscourseConnectBase.sign(redirect_query["sso"][0], "newSecretForOverRainbow")
expect(redirect_query["sig"][0]).to eq(expected_sig)
get "/session/sso_provider",
params: Rack::Utils.parse_query(@sso.payload("oldSecretForOverRainbow"))
expect(response.status).to eq(302)
redirect_uri = URI.parse(response.location)
expect(redirect_uri.host).to eq("somewhere.over.rainbow")
redirect_query = CGI.parse(redirect_uri.query)
expected_sig =
DiscourseConnectBase.sign(redirect_query["sso"][0], "oldSecretForOverRainbow")
expect(redirect_query["sig"][0]).to eq(expected_sig)
end
it "it fails to log in if secret is wrong" do
get "/session/sso_provider",
params: Rack::Utils.parse_query(@sso.payload("secretForRandomSite"))
expect(response.status).to eq(422)
end
it "fails with a nice error message if secret is blank" do
FEATURE: Rename 'Discourse SSO' to DiscourseConnect (#11978) The 'Discourse SSO' protocol is being rebranded to DiscourseConnect. This should help to reduce confusion when 'SSO' is used in the generic sense. This commit aims to: - Rename `sso_` site settings. DiscourseConnect specific ones are prefixed `discourse_connect_`. Generic settings are prefixed `auth_` - Add (server-side-only) backwards compatibility for the old setting names, with deprecation notices - Copy `site_settings` database records to the new names - Rename relevant translation keys - Update relevant translations This commit does **not** aim to: - Rename any Ruby classes or methods. This might be done in a future commit - Change any URLs. This would break existing integrations - Make any changes to the protocol. This would break existing integrations - Change any functionality. Further normalization across DiscourseConnect and other auth methods will be done separately The risks are: - There is no backwards compatibility for site settings on the client-side. Accessing auth-related site settings in Javascript is fairly rare, and an error on the client side would not be security-critical. - If a plugin is monkey-patching parts of the auth process, changes to locale keys could cause broken error messages. This should also be unlikely. The old site setting names remain functional, so security-related overrides will remain working. A follow-up commit will be made with a post-deploy migration to delete the old `site_settings` rows.
2021-02-08 18:04:33 +08:00
SiteSetting.discourse_connect_provider_secrets = ""
sso = DiscourseConnectProvider.new
sso.nonce = "mynonce"
sso.return_sso_url = "http://website.without.secret.com/sso"
get "/session/sso_provider", params: Rack::Utils.parse_query(sso.payload("aasdasdasd"))
expect(response.status).to eq(400)
FEATURE: Rename 'Discourse SSO' to DiscourseConnect (#11978) The 'Discourse SSO' protocol is being rebranded to DiscourseConnect. This should help to reduce confusion when 'SSO' is used in the generic sense. This commit aims to: - Rename `sso_` site settings. DiscourseConnect specific ones are prefixed `discourse_connect_`. Generic settings are prefixed `auth_` - Add (server-side-only) backwards compatibility for the old setting names, with deprecation notices - Copy `site_settings` database records to the new names - Rename relevant translation keys - Update relevant translations This commit does **not** aim to: - Rename any Ruby classes or methods. This might be done in a future commit - Change any URLs. This would break existing integrations - Make any changes to the protocol. This would break existing integrations - Change any functionality. Further normalization across DiscourseConnect and other auth methods will be done separately The risks are: - There is no backwards compatibility for site settings on the client-side. Accessing auth-related site settings in Javascript is fairly rare, and an error on the client side would not be security-critical. - If a plugin is monkey-patching parts of the auth process, changes to locale keys could cause broken error messages. This should also be unlikely. The old site setting names remain functional, so security-related overrides will remain working. A follow-up commit will be made with a post-deploy migration to delete the old `site_settings` rows.
2021-02-08 18:04:33 +08:00
expect(response.body).to eq(I18n.t("discourse_connect.missing_secret"))
end
it "returns a 422 if no return_sso_url" do
FEATURE: Rename 'Discourse SSO' to DiscourseConnect (#11978) The 'Discourse SSO' protocol is being rebranded to DiscourseConnect. This should help to reduce confusion when 'SSO' is used in the generic sense. This commit aims to: - Rename `sso_` site settings. DiscourseConnect specific ones are prefixed `discourse_connect_`. Generic settings are prefixed `auth_` - Add (server-side-only) backwards compatibility for the old setting names, with deprecation notices - Copy `site_settings` database records to the new names - Rename relevant translation keys - Update relevant translations This commit does **not** aim to: - Rename any Ruby classes or methods. This might be done in a future commit - Change any URLs. This would break existing integrations - Make any changes to the protocol. This would break existing integrations - Change any functionality. Further normalization across DiscourseConnect and other auth methods will be done separately The risks are: - There is no backwards compatibility for site settings on the client-side. Accessing auth-related site settings in Javascript is fairly rare, and an error on the client side would not be security-critical. - If a plugin is monkey-patching parts of the auth process, changes to locale keys could cause broken error messages. This should also be unlikely. The old site setting names remain functional, so security-related overrides will remain working. A follow-up commit will be made with a post-deploy migration to delete the old `site_settings` rows.
2021-02-08 18:04:33 +08:00
SiteSetting.discourse_connect_provider_secrets = "abcdefghij"
get "/session/sso_provider?sso=asdf&sig=abcdefghij"
expect(response.status).to eq(422)
end
it "successfully redirects user to return_sso_url when the user is logged in" do
sign_in(@user)
get "/session/sso_provider",
params: Rack::Utils.parse_query(@sso.payload("secretForOverRainbow"))
location = response.header["Location"]
expect(location).to match(%r{^http://somewhere.over.rainbow/sso})
payload = location.split("?")[1]
sso2 = DiscourseConnectProvider.parse(payload)
expect(sso2.email).to eq(@user.email)
expect(sso2.name).to eq(@user.name)
expect(sso2.username).to eq(@user.username)
expect(sso2.external_id).to eq(@user.id.to_s)
expect(sso2.admin).to eq(true)
expect(sso2.moderator).to eq(false)
expect(sso2.groups).to eq(@user.groups.pluck(:name).join(","))
expect(sso2.avatar_url.blank?).to_not eq(true)
expect(sso2.profile_background_url.blank?).to_not eq(true)
expect(sso2.card_background_url.blank?).to_not eq(true)
expect(sso2.avatar_url).to start_with(Discourse.base_url)
expect(sso2.profile_background_url).to start_with(Discourse.base_url)
expect(sso2.card_background_url).to start_with(Discourse.base_url)
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
expect(sso2.confirmed_2fa).to eq(nil)
expect(sso2.no_2fa_methods).to eq(nil)
end
it "handles non local content correctly" do
SiteSetting.avatar_sizes = "100|49"
setup_s3
SiteSetting.s3_cdn_url = "http://cdn.com"
stub_request(:any, /s3-upload-bucket.s3.dualstack.us-west-1.amazonaws.com/).to_return(
status: 200,
body: "",
headers: {
referer: "fgdfds",
},
)
@user.create_user_avatar!
upload =
Fabricate(
:upload,
url: "//s3-upload-bucket.s3.dualstack.us-west-1.amazonaws.com/something",
)
Fabricate(
:optimized_image,
sha1: SecureRandom.hex << "A" * 8,
upload: upload,
width: 98,
height: 98,
url: "//s3-upload-bucket.s3.amazonaws.com/something/else",
)
@user.update_columns(uploaded_avatar_id: upload.id)
upload1 = Fabricate(:upload_s3)
upload2 = Fabricate(:upload_s3)
@user.user_profile.update!(
profile_background_upload: upload1,
card_background_upload: upload2,
)
@user.reload
@user.user_avatar.reload
@user.user_profile.reload
sign_in(@user)
stub_request(:get, "http://cdn.com/something/else").to_return(
body: lambda { |request| File.new(Rails.root + "spec/fixtures/images/logo.png") },
)
get "/session/sso_provider",
params: Rack::Utils.parse_query(@sso.payload("secretForOverRainbow"))
location = response.header["Location"]
# javascript code will handle redirection of user to return_sso_url
expect(location).to match(%r{^http://somewhere.over.rainbow/sso})
payload = location.split("?")[1]
sso2 = DiscourseConnectProvider.parse(payload)
expect(sso2.avatar_url.blank?).to_not eq(true)
expect(sso2.profile_background_url.blank?).to_not eq(true)
expect(sso2.card_background_url.blank?).to_not eq(true)
expect(sso2.avatar_url).to start_with("#{SiteSetting.s3_cdn_url}/original")
expect(sso2.profile_background_url).to start_with(SiteSetting.s3_cdn_url)
expect(sso2.card_background_url).to start_with(SiteSetting.s3_cdn_url)
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
expect(sso2.confirmed_2fa).to eq(nil)
expect(sso2.no_2fa_methods).to eq(nil)
end
it "successfully logs out and redirects user to return_sso_url when the user is logged in" do
sign_in(@user)
@sso.logout = true
get "/session/sso_provider",
params: Rack::Utils.parse_query(@sso.payload("secretForOverRainbow"))
location = response.header["Location"]
expect(location).to match(%r{^http://somewhere.over.rainbow/sso$})
expect(response.status).to eq(302)
expect(session[:current_user_id]).to be_blank
expect(response.cookies["_t"]).to be_blank
end
it "successfully logs out and redirects user to return_sso_url when the user is not logged in" do
@sso.logout = true
get "/session/sso_provider",
params: Rack::Utils.parse_query(@sso.payload("secretForOverRainbow"))
location = response.header["Location"]
expect(location).to match(%r{^http://somewhere.over.rainbow/sso$})
expect(response.status).to eq(302)
expect(session[:current_user_id]).to be_blank
expect(response.cookies["_t"]).to be_blank
end
end
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
describe "can act as a 2FA provider" do
fab!(:user_totp) { Fabricate(:user_second_factor_totp, user: user) }
before { @sso.require_2fa = true }
it "requires the user to confirm 2FA before they are redirected to the SSO return URL" do
sign_in(user)
get "/session/sso_provider",
params: Rack::Utils.parse_query(@sso.payload("secretForOverRainbow"))
uri = URI(response.location)
expect(uri.hostname).to eq(Discourse.current_hostname)
expect(uri.path).to eq("/session/2fa")
nonce = uri.query.match(/\Anonce=([A-Za-z0-9]{32})\Z/)[1]
expect(nonce).to be_present
# attempt no. 1 to bypass 2fa
get "/session/sso_provider", params: { second_factor_nonce: nonce }
expect(response.status).to eq(401)
expect(response.parsed_body["error"]).to eq(
I18n.t("second_factor_auth.challenge_not_completed"),
)
# attempt no. 2 to bypass 2fa
get "/session/sso_provider",
params: { second_factor_nonce: nonce }.merge(
Rack::Utils.parse_query(@sso.payload("secretForOverRainbow")),
)
expect(response.status).to eq(401)
expect(response.parsed_body["error"]).to eq(
I18n.t("second_factor_auth.challenge_not_completed"),
)
# confirm 2fa
post "/session/2fa.json",
params: {
nonce: nonce,
second_factor_token: ROTP::TOTP.new(user_totp.data).now,
second_factor_method: UserSecondFactor.methods[:totp],
}
expect(response.status).to eq(200)
expect(response.parsed_body["ok"]).to eq(true)
expect(response.parsed_body["callback_method"]).to eq("GET")
expect(response.parsed_body["callback_path"]).to eq("/session/sso_provider")
expect(response.parsed_body["redirect_url"]).to be_blank
get "/session/sso_provider", params: { second_factor_nonce: nonce }
expect(response.status).to eq(200)
expect(response.parsed_body["success"]).to eq("OK")
redirect_url = response.parsed_body["redirect_url"]
expect(redirect_url).to start_with("http://somewhere.over.rainbow/sso?sso=")
sso = DiscourseConnectProvider.parse(URI(redirect_url).query)
expect(sso.confirmed_2fa).to eq(true)
expect(sso.no_2fa_methods).to eq(nil)
expect(sso.username).to eq(user.username)
expect(sso.email).to eq(user.email)
end
it "doesn't accept backup codes" do
backup_codes = user.generate_backup_codes
sign_in(user)
get "/session/sso_provider",
params: Rack::Utils.parse_query(@sso.payload("secretForOverRainbow"))
uri = URI(response.location)
expect(uri.hostname).to eq(Discourse.current_hostname)
expect(uri.path).to eq("/session/2fa")
nonce = uri.query.match(/\Anonce=([A-Za-z0-9]{32})\Z/)[1]
expect(nonce).to be_present
post "/session/2fa.json",
params: {
nonce: nonce,
second_factor_token: backup_codes.sample,
second_factor_method: UserSecondFactor.methods[:backup_codes],
}
expect(response.status).to eq(403)
get "/session/sso_provider", params: { second_factor_nonce: nonce }
expect(response.status).to eq(401)
expect(response.parsed_body["error"]).to eq(
I18n.t("second_factor_auth.challenge_not_completed"),
)
end
context "when the user has no 2fa methods" do
before do
user_totp.destroy!
user.reload
end
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
it "redirects the user back to the SSO return url and indicates in the payload that they do not have 2fa methods" do
sign_in(user)
get "/session/sso_provider",
params: Rack::Utils.parse_query(@sso.payload("secretForOverRainbow"))
expect(response.status).to eq(302)
redirect_url = response.location
expect(redirect_url).to start_with("http://somewhere.over.rainbow/sso?sso=")
sso = DiscourseConnectProvider.parse(URI(redirect_url).query)
expect(sso.confirmed_2fa).to eq(nil)
expect(sso.no_2fa_methods).to eq(true)
expect(sso.username).to eq(user.username)
expect(sso.email).to eq(user.email)
end
end
context "when there is no logged in user" do
it "redirects the user to login first" do
get "/session/sso_provider",
params: Rack::Utils.parse_query(@sso.payload("secretForOverRainbow"))
expect(response.status).to eq(302)
expect(response.location).to eq("http://#{Discourse.current_hostname}/login")
end
it "doesn't make the user confirm 2fa twice if they've just logged in and confirmed 2fa while doing so" do
get "/session/sso_provider",
params: Rack::Utils.parse_query(@sso.payload("secretForOverRainbow"))
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
post "/session.json",
params: {
login: user.username,
password: "myfrogs123ADMIN",
second_factor_token: ROTP::TOTP.new(user_totp.data).now,
second_factor_method: UserSecondFactor.methods[:totp],
},
xhr: true,
headers: headers
expect(response.status).to eq(204)
# the frontend will take care of actually redirecting the user
redirect_url = response.cookies["sso_destination_url"]
expect(redirect_url).to start_with("http://somewhere.over.rainbow/sso?sso=")
sso = DiscourseConnectProvider.parse(URI(redirect_url).query)
expect(sso.confirmed_2fa).to eq(true)
expect(sso.no_2fa_methods).to eq(nil)
expect(sso.username).to eq(user.username)
expect(sso.email).to eq(user.email)
end
it "doesn't indicate the user has confirmed 2fa after they've logged in if they have no 2fa methods" do
user_totp.destroy!
user.reload
get "/session/sso_provider",
params: Rack::Utils.parse_query(@sso.payload("secretForOverRainbow"))
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
post "/session.json",
params: {
login: user.username,
password: "myfrogs123ADMIN",
},
xhr: true,
headers: headers
redirect_url = response.cookies["sso_destination_url"]
expect(redirect_url).to start_with("http://somewhere.over.rainbow/sso?sso=")
sso = DiscourseConnectProvider.parse(URI(redirect_url).query)
expect(sso.confirmed_2fa).to eq(nil)
expect(sso.no_2fa_methods).to eq(true)
expect(sso.username).to eq(user.username)
expect(sso.email).to eq(user.email)
end
end
end
end
describe "#create" do
context "when read only mode" do
use_redis_snapshotting
before do
Discourse.enable_readonly_mode
EmailToken.confirm(email_token.token)
EmailToken.confirm(admin_email_token.token)
end
it "prevents login by regular users" do
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
expect(response.status).not_to eq(200)
end
it "prevents login by admins" do
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
expect(response.status).not_to eq(200)
end
end
context "when in staff writes only mode" do
use_redis_snapshotting
before do
Discourse.enable_readonly_mode(Discourse::STAFF_WRITES_ONLY_MODE_KEY)
EmailToken.confirm(email_token.token)
EmailToken.confirm(admin_email_token.token)
end
it "allows admin login" do
post "/session.json", params: { login: admin.username, password: "myawesomepassword" }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
end
it "prevents login by regular users" do
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
expect(response.status).not_to eq(200)
end
end
context "when local login is disabled" do
before do
SiteSetting.enable_local_logins = false
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
end
it_behaves_like "failed to continue local login"
end
context "when SSO is enabled" do
before do
FEATURE: Rename 'Discourse SSO' to DiscourseConnect (#11978) The 'Discourse SSO' protocol is being rebranded to DiscourseConnect. This should help to reduce confusion when 'SSO' is used in the generic sense. This commit aims to: - Rename `sso_` site settings. DiscourseConnect specific ones are prefixed `discourse_connect_`. Generic settings are prefixed `auth_` - Add (server-side-only) backwards compatibility for the old setting names, with deprecation notices - Copy `site_settings` database records to the new names - Rename relevant translation keys - Update relevant translations This commit does **not** aim to: - Rename any Ruby classes or methods. This might be done in a future commit - Change any URLs. This would break existing integrations - Make any changes to the protocol. This would break existing integrations - Change any functionality. Further normalization across DiscourseConnect and other auth methods will be done separately The risks are: - There is no backwards compatibility for site settings on the client-side. Accessing auth-related site settings in Javascript is fairly rare, and an error on the client side would not be security-critical. - If a plugin is monkey-patching parts of the auth process, changes to locale keys could cause broken error messages. This should also be unlikely. The old site setting names remain functional, so security-related overrides will remain working. A follow-up commit will be made with a post-deploy migration to delete the old `site_settings` rows.
2021-02-08 18:04:33 +08:00
SiteSetting.discourse_connect_url = "https://www.example.com/sso"
SiteSetting.enable_discourse_connect = true
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
end
it_behaves_like "failed to continue local login"
end
context "when local login via email is disabled" do
before do
SiteSetting.enable_local_logins_via_email = false
EmailToken.confirm(email_token.token)
end
it "doesnt matter, logs in correctly" do
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
end
end
context "when email is confirmed" do
before { EmailToken.confirm(email_token.token) }
it "raises an error when the login isn't present" do
post "/session.json"
expect(response.status).to eq(400)
end
describe "invalid password" do
it "should return an error with an invalid password" do
post "/session.json", params: { login: user.username, password: "sssss" }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(
I18n.t("login.incorrect_username_email_or_password"),
)
end
end
describe "invalid password" do
it "should return an error with an invalid password if too long" do
User.any_instance.expects(:confirm_password?).never
post "/session.json",
params: {
login: user.username,
password: ("s" * (User.max_password_length + 1)),
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(
I18n.t("login.incorrect_username_email_or_password"),
)
end
end
describe "suspended user" do
it "should return an error" do
user.suspended_till = 2.days.from_now
user.suspended_at = Time.now
user.save!
StaffActionLogger.new(user).log_user_suspend(user, "<strike>banned</strike>")
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
expected_message =
I18n.t(
"login.suspended_with_reason",
date: I18n.l(user.suspended_till, format: :date_only),
reason: Rack::Utils.escape_html(user.suspend_reason),
)
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(expected_message)
end
it "when suspended forever should return an error without suspended till date" do
user.suspended_till = 101.years.from_now
user.suspended_at = Time.now
user.save!
StaffActionLogger.new(user).log_user_suspend(user, "<strike>banned</strike>")
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
expected_message =
I18n.t(
"login.suspended_with_reason_forever",
reason: Rack::Utils.escape_html(user.suspend_reason),
)
expect(response.parsed_body["error"]).to eq(expected_message)
end
end
describe "deactivated user" do
it "should return an error" do
user.active = false
user.save!
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(I18n.t("login.not_activated"))
end
end
describe "success by username" do
it "logs in correctly" do
events =
DiscourseEvent.track_events do
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
end
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(events.map { |event| event[:event_name] }).to contain_exactly(
:user_logged_in,
:user_first_logged_in,
)
user.reload
expect(session[:current_user_id]).to eq(user.id)
expect(user.user_auth_tokens.count).to eq(1)
FEATURE: Apply rate limits per user instead of IP for trusted users (#14706) Currently, Discourse rate limits all incoming requests by the IP address they originate from regardless of the user making the request. This can be frustrating if there are multiple users using Discourse simultaneously while sharing the same IP address (e.g. employees in an office). This commit implements a new feature to make Discourse apply rate limits by user id rather than IP address for users at or higher than the configured trust level (1 is the default). For example, let's say a Discourse instance is configured to allow 200 requests per minute per IP address, and we have 10 users at trust level 4 using Discourse simultaneously from the same IP address. Before this feature, the 10 users could only make a total of 200 requests per minute before they got rate limited. But with the new feature, each user is allowed to make 200 requests per minute because the rate limits are applied on user id rather than the IP address. The minimum trust level for applying user-id-based rate limits can be configured by the `skip_per_ip_rate_limit_trust_level` global setting. The default is 1, but it can be changed by either adding the `DISCOURSE_SKIP_PER_IP_RATE_LIMIT_TRUST_LEVEL` environment variable with the desired value to your `app.yml`, or changing the setting's value in the `discourse.conf` file. Requests made with API keys are still rate limited by IP address and the relevant global settings that control API keys rate limits. Before this commit, Discourse's auth cookie (`_t`) was simply a 32 characters string that Discourse used to lookup the current user from the database and the cookie contained no additional information about the user. However, we had to change the cookie content in this commit so we could identify the user from the cookie without making a database query before the rate limits logic and avoid introducing a bottleneck on busy sites. Besides the 32 characters auth token, the cookie now includes the user id, trust level and the cookie's generation date, and we encrypt/sign the cookie to prevent tampering. Internal ticket number: t54739.
2021-11-18 04:27:30 +08:00
unhashed_token = decrypt_auth_cookie(cookies[:_t])[:token]
expect(UserAuthToken.hash_token(unhashed_token)).to eq(
user.user_auth_tokens.first.auth_token,
)
end
context "when timezone param is provided" do
it "sets the user_option timezone for the user" do
post "/session.json",
params: {
login: user.username,
password: "myawesomepassword",
timezone: "Australia/Melbourne",
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(user.reload.user_option.timezone).to eq("Australia/Melbourne")
end
end
end
context "when a user has security key-only 2FA login" do
let!(:user_security_key) do
Fabricate(
:user_security_key,
user: user,
credential_id: valid_security_key_data[:credential_id],
public_key: valid_security_key_data[:public_key],
)
end
before do
simulate_localhost_webauthn_challenge
# store challenge in secure session by failing login once
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
end
context "when the security key params are blank and a random second factor token is provided" do
it "shows an error message and denies login" do
post "/session.json",
params: {
login: user.username,
password: "myawesomepassword",
second_factor_token: "99999999",
2020-01-15 18:27:12 +08:00
second_factor_method: UserSecondFactor.methods[:security_key],
}
expect(response.status).to eq(200)
expect(session[:current_user_id]).to eq(nil)
response_body = response.parsed_body
expect(response_body["failed"]).to eq("FAILED")
expect(response_body["error"]).to eq(I18n.t("login.invalid_security_key"))
end
end
context "when the security key params are invalid" do
it "shows an error message and denies login" do
post "/session.json",
params: {
login: user.username,
password: "myawesomepassword",
2020-01-15 18:27:12 +08:00
second_factor_token: {
signature: "bad_sig",
clientData: "bad_clientData",
credentialId: "bad_credential_id",
authenticatorData: "bad_authenticator_data",
},
second_factor_method: UserSecondFactor.methods[:security_key],
}
expect(response.status).to eq(200)
expect(session[:current_user_id]).to eq(nil)
response_body = response.parsed_body
expect(response_body["failed"]).to eq("FAILED")
expect(response_body["error"]).to eq(I18n.t("webauthn.validation.not_found_error"))
end
end
context "when the security key params are valid" do
it "logs the user in" do
post "/session.json",
params: {
login: user.username,
password: "myawesomepassword",
2020-01-15 18:27:12 +08:00
second_factor_token: valid_security_key_auth_post_data,
second_factor_method: UserSecondFactor.methods[:security_key],
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
user.reload
expect(session[:current_user_id]).to eq(user.id)
expect(user.user_auth_tokens.count).to eq(1)
end
end
context "when the security key is disabled in the background by the user and TOTP is enabled" do
before do
user_security_key.destroy!
Fabricate(:user_second_factor_totp, user: user)
end
it "shows an error message and denies login" do
post "/session.json",
params: {
login: user.username,
password: "myawesomepassword",
2020-01-15 18:27:12 +08:00
second_factor_token: valid_security_key_auth_post_data,
second_factor_method: UserSecondFactor.methods[:security_key],
}
expect(response.status).to eq(200)
expect(session[:current_user_id]).to eq(nil)
response_body = response.parsed_body
expect(response_body["failed"]).to eq("FAILED")
expect(response.parsed_body["error"]).to eq(
2020-01-15 18:27:12 +08:00
I18n.t("login.not_enabled_second_factor_method"),
)
end
end
end
context "when user has TOTP-only 2FA login" do
2018-06-28 16:12:32 +08:00
let!(:user_second_factor) { Fabricate(:user_second_factor_totp, user: user) }
let!(:user_second_factor_backup) { Fabricate(:user_second_factor_backup, user: user) }
describe "when second factor token is missing" do
it "should return the right response" do
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(
2020-01-15 18:27:12 +08:00
I18n.t("login.invalid_second_factor_method"),
)
end
end
describe "when second factor token is invalid" do
2018-06-28 16:12:32 +08:00
context "when using totp method" do
it "should return the right response" do
post "/session.json",
params: {
login: user.username,
password: "myawesomepassword",
second_factor_token: "00000000",
second_factor_method: UserSecondFactor.methods[:totp],
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(
2018-06-28 16:12:32 +08:00
I18n.t("login.invalid_second_factor_code"),
)
end
end
2018-06-28 16:12:32 +08:00
context "when using backup code method" do
it "should return the right response" do
post "/session.json",
params: {
login: user.username,
password: "myawesomepassword",
second_factor_token: "00000000",
second_factor_method: UserSecondFactor.methods[:backup_codes],
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(
2018-06-28 16:12:32 +08:00
I18n.t("login.invalid_second_factor_code"),
)
end
end
end
describe "when second factor token is valid" do
2018-06-28 16:12:32 +08:00
context "when using totp method" do
it "should log the user in" do
post "/session.json",
params: {
login: user.username,
password: "myawesomepassword",
second_factor_token: ROTP::TOTP.new(user_second_factor.data).now,
second_factor_method: UserSecondFactor.methods[:totp],
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
2018-06-28 16:12:32 +08:00
user.reload
expect(session[:current_user_id]).to eq(user.id)
expect(user.user_auth_tokens.count).to eq(1)
FEATURE: Apply rate limits per user instead of IP for trusted users (#14706) Currently, Discourse rate limits all incoming requests by the IP address they originate from regardless of the user making the request. This can be frustrating if there are multiple users using Discourse simultaneously while sharing the same IP address (e.g. employees in an office). This commit implements a new feature to make Discourse apply rate limits by user id rather than IP address for users at or higher than the configured trust level (1 is the default). For example, let's say a Discourse instance is configured to allow 200 requests per minute per IP address, and we have 10 users at trust level 4 using Discourse simultaneously from the same IP address. Before this feature, the 10 users could only make a total of 200 requests per minute before they got rate limited. But with the new feature, each user is allowed to make 200 requests per minute because the rate limits are applied on user id rather than the IP address. The minimum trust level for applying user-id-based rate limits can be configured by the `skip_per_ip_rate_limit_trust_level` global setting. The default is 1, but it can be changed by either adding the `DISCOURSE_SKIP_PER_IP_RATE_LIMIT_TRUST_LEVEL` environment variable with the desired value to your `app.yml`, or changing the setting's value in the `discourse.conf` file. Requests made with API keys are still rate limited by IP address and the relevant global settings that control API keys rate limits. Before this commit, Discourse's auth cookie (`_t`) was simply a 32 characters string that Discourse used to lookup the current user from the database and the cookie contained no additional information about the user. However, we had to change the cookie content in this commit so we could identify the user from the cookie without making a database query before the rate limits logic and avoid introducing a bottleneck on busy sites. Besides the 32 characters auth token, the cookie now includes the user id, trust level and the cookie's generation date, and we encrypt/sign the cookie to prevent tampering. Internal ticket number: t54739.
2021-11-18 04:27:30 +08:00
unhashed_token = decrypt_auth_cookie(cookies[:_t])[:token]
expect(UserAuthToken.hash_token(unhashed_token)).to eq(
2018-06-28 16:12:32 +08:00
user.user_auth_tokens.first.auth_token,
)
end
end
2018-06-28 16:12:32 +08:00
context "when using backup code method" do
it "should log the user in" do
post "/session.json",
params: {
login: user.username,
password: "myawesomepassword",
second_factor_token: "iAmValidBackupCode",
second_factor_method: UserSecondFactor.methods[:backup_codes],
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
2018-06-28 16:12:32 +08:00
user.reload
expect(session[:current_user_id]).to eq(user.id)
expect(user.user_auth_tokens.count).to eq(1)
FEATURE: Apply rate limits per user instead of IP for trusted users (#14706) Currently, Discourse rate limits all incoming requests by the IP address they originate from regardless of the user making the request. This can be frustrating if there are multiple users using Discourse simultaneously while sharing the same IP address (e.g. employees in an office). This commit implements a new feature to make Discourse apply rate limits by user id rather than IP address for users at or higher than the configured trust level (1 is the default). For example, let's say a Discourse instance is configured to allow 200 requests per minute per IP address, and we have 10 users at trust level 4 using Discourse simultaneously from the same IP address. Before this feature, the 10 users could only make a total of 200 requests per minute before they got rate limited. But with the new feature, each user is allowed to make 200 requests per minute because the rate limits are applied on user id rather than the IP address. The minimum trust level for applying user-id-based rate limits can be configured by the `skip_per_ip_rate_limit_trust_level` global setting. The default is 1, but it can be changed by either adding the `DISCOURSE_SKIP_PER_IP_RATE_LIMIT_TRUST_LEVEL` environment variable with the desired value to your `app.yml`, or changing the setting's value in the `discourse.conf` file. Requests made with API keys are still rate limited by IP address and the relevant global settings that control API keys rate limits. Before this commit, Discourse's auth cookie (`_t`) was simply a 32 characters string that Discourse used to lookup the current user from the database and the cookie contained no additional information about the user. However, we had to change the cookie content in this commit so we could identify the user from the cookie without making a database query before the rate limits logic and avoid introducing a bottleneck on busy sites. Besides the 32 characters auth token, the cookie now includes the user id, trust level and the cookie's generation date, and we encrypt/sign the cookie to prevent tampering. Internal ticket number: t54739.
2021-11-18 04:27:30 +08:00
unhashed_token = decrypt_auth_cookie(cookies[:_t])[:token]
expect(UserAuthToken.hash_token(unhashed_token)).to eq(
2018-06-28 16:12:32 +08:00
user.user_auth_tokens.first.auth_token,
)
end
end
end
end
describe "with a blocked IP" do
it "doesn't log in" do
ScreenedIpAddress.all.destroy_all
get "/"
_screened_ip = Fabricate(:screened_ip_address, ip_address: request.remote_ip)
post "/session.json",
params: {
login: "@" + user.username,
password: "myawesomepassword",
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to be_present
user.reload
expect(session[:current_user_id]).to be_nil
end
end
describe "strips leading @ symbol" do
it "sets a session id" do
post "/session.json",
params: {
login: "@" + user.username,
password: "myawesomepassword",
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
user.reload
expect(session[:current_user_id]).to eq(user.id)
end
end
describe "also allow login by email" do
it "sets a session id" do
post "/session.json", params: { login: user.email, password: "myawesomepassword" }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(session[:current_user_id]).to eq(user.id)
end
end
context "when login has leading and trailing space" do
let(:username) { " #{user.username} " }
let(:email) { " #{user.email} " }
it "strips spaces from the username" do
post "/session.json", params: { login: username, password: "myawesomepassword" }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
end
it "strips spaces from the email" do
post "/session.json", params: { login: email, password: "myawesomepassword" }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
end
end
describe "when the site requires approval of users" do
before { SiteSetting.must_approve_users = true }
context "with an unapproved user" do
before do
user.update_columns(approved: false)
post "/session.json", params: { login: user.email, password: "myawesomepassword" }
end
it "doesn't log in the user" do
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to be_present
expect(session[:current_user_id]).to be_blank
end
it "shows the 'not approved' error message" do
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(I18n.t("login.not_approved"))
end
end
context "with an unapproved user who is an admin" do
it "sets a session id" do
user.admin = true
user.save!
post "/session.json", params: { login: user.email, password: "myawesomepassword" }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(session[:current_user_id]).to eq(user.id)
end
end
end
context "when admins are restricted by ip address" do
before do
SiteSetting.use_admin_ip_allowlist = true
ScreenedIpAddress.all.destroy_all
end
it "is successful for admin at the ip address" do
get "/"
Fabricate(
:screened_ip_address,
ip_address: request.remote_ip,
action_type: ScreenedIpAddress.actions[:allow_admin],
)
user.admin = true
user.save!
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(session[:current_user_id]).to eq(user.id)
end
it "returns an error for admin not at the ip address" do
Fabricate(
:screened_ip_address,
ip_address: "111.234.23.11",
action_type: ScreenedIpAddress.actions[:allow_admin],
)
user.admin = true
user.save!
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to be_present
expect(session[:current_user_id]).not_to eq(user.id)
end
it "is successful for non-admin not at the ip address" do
Fabricate(
:screened_ip_address,
ip_address: "111.234.23.11",
action_type: ScreenedIpAddress.actions[:allow_admin],
)
user.admin = false
user.save!
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(session[:current_user_id]).to eq(user.id)
end
end
end
context "when email has not been confirmed" do
def post_login
post "/session.json", params: { login: user.email, password: "myawesomepassword" }
end
it "doesn't log in the user" do
post_login
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to be_present
expect(session[:current_user_id]).to be_blank
end
it "shows the 'not activated' error message" do
post_login
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(I18n.t "login.not_activated")
end
context "when the 'must approve users' site setting is enabled" do
before { SiteSetting.must_approve_users = true }
it "shows the 'not approved' error message" do
post_login
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to eq(I18n.t "login.not_approved")
end
end
end
context "when rate limited" do
it "rate limits login" do
SiteSetting.max_logins_per_ip_per_hour = 2
RateLimiter.enable
RateLimiter.clear_all!
EmailToken.confirm(email_token.token)
2.times do
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
end
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
expect(response.status).to eq(429)
json = response.parsed_body
expect(json["error_type"]).to eq("rate_limit")
end
it "rate limits second factor attempts by IP" do
RateLimiter.enable
RateLimiter.clear_all!
6.times do |x|
post "/session.json",
params: {
login: "#{user.username}#{x}",
password: "myawesomepassword",
second_factor_token: "000000",
second_factor_method: UserSecondFactor.methods[:totp],
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).to be_present
end
post "/session.json",
params: {
login: user.username,
password: "myawesomepassword",
second_factor_token: "000000",
second_factor_method: UserSecondFactor.methods[:totp],
}
expect(response.status).to eq(429)
json = response.parsed_body
expect(json["error_type"]).to eq("rate_limit")
end
it "rate limits second factor attempts by login" do
RateLimiter.enable
RateLimiter.clear_all!
EmailToken.confirm(email_token.token)
6.times do |x|
post "/session.json",
params: {
login: user.username,
password: "myawesomepassword",
second_factor_token: "000000",
second_factor_method: UserSecondFactor.methods[:totp],
},
env: {
REMOTE_ADDR: "1.2.3.#{x}",
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
end
[
user.username + " ",
user.username.capitalize,
user.username,
].each_with_index do |username, x|
post "/session.json",
params: {
login: username,
password: "myawesomepassword",
second_factor_token: "000000",
second_factor_method: UserSecondFactor.methods[:totp],
},
env: {
REMOTE_ADDR: "1.2.4.#{x}",
}
expect(response.status).to eq(429)
json = response.parsed_body
expect(json["error_type"]).to eq("rate_limit")
end
end
end
end
describe "#destroy" do
it "removes the session variable and the auth token cookies" do
user = sign_in(Fabricate(:user))
delete "/session/#{user.username}.json"
expect(response.status).to eq(302)
expect(session[:current_user_id]).to be_blank
expect(response.cookies["_t"]).to be_blank
end
it "returns the redirect URL in the body for XHR requests" do
user = sign_in(Fabricate(:user))
delete "/session/#{user.username}.json", xhr: true
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(session[:current_user_id]).to be_blank
expect(response.cookies["_t"]).to be_blank
expect(response.parsed_body["redirect_url"]).to eq("/")
end
it "redirects to /login when SSO and login_required" do
FEATURE: Rename 'Discourse SSO' to DiscourseConnect (#11978) The 'Discourse SSO' protocol is being rebranded to DiscourseConnect. This should help to reduce confusion when 'SSO' is used in the generic sense. This commit aims to: - Rename `sso_` site settings. DiscourseConnect specific ones are prefixed `discourse_connect_`. Generic settings are prefixed `auth_` - Add (server-side-only) backwards compatibility for the old setting names, with deprecation notices - Copy `site_settings` database records to the new names - Rename relevant translation keys - Update relevant translations This commit does **not** aim to: - Rename any Ruby classes or methods. This might be done in a future commit - Change any URLs. This would break existing integrations - Make any changes to the protocol. This would break existing integrations - Change any functionality. Further normalization across DiscourseConnect and other auth methods will be done separately The risks are: - There is no backwards compatibility for site settings on the client-side. Accessing auth-related site settings in Javascript is fairly rare, and an error on the client side would not be security-critical. - If a plugin is monkey-patching parts of the auth process, changes to locale keys could cause broken error messages. This should also be unlikely. The old site setting names remain functional, so security-related overrides will remain working. A follow-up commit will be made with a post-deploy migration to delete the old `site_settings` rows.
2021-02-08 18:04:33 +08:00
SiteSetting.discourse_connect_url = "https://example.com/sso"
SiteSetting.enable_discourse_connect = true
user = sign_in(Fabricate(:user))
delete "/session/#{user.username}.json", xhr: true
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(response.parsed_body["redirect_url"]).to eq("/")
SiteSetting.login_required = true
user = sign_in(Fabricate(:user))
delete "/session/#{user.username}.json", xhr: true
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(response.parsed_body["redirect_url"]).to eq("/login")
end
it "allows plugins to manipulate redirect URL" do
callback = ->(data) { data[:redirect_url] = "/myredirect/#{data[:user].username}" }
DiscourseEvent.on(:before_session_destroy, &callback)
user = sign_in(Fabricate(:user))
delete "/session/#{user.username}.json", xhr: true
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(response.parsed_body["redirect_url"]).to eq("/myredirect/#{user.username}")
ensure
DiscourseEvent.off(:before_session_destroy, &callback)
end
it "includes ip and user agent in the before_session_destroy event params" do
callback_params = {}
callback = ->(data) { callback_params = data }
DiscourseEvent.on(:before_session_destroy, &callback)
user = sign_in(Fabricate(:user))
delete "/session/#{user.username}.json",
xhr: true,
headers: {
HTTP_USER_AGENT: "AwesomeBrowser",
}
expect(callback_params[:user_agent]).to eq("AwesomeBrowser")
expect(callback_params[:client_ip]).to eq("127.0.0.1")
ensure
DiscourseEvent.off(:before_session_destroy, &callback)
end
end
describe "#one_time_password" do
context "with missing token" do
it "returns the right response" do
get "/session/otp"
expect(response.status).to eq(404)
end
end
context "with invalid token" do
it "returns the right response" do
get "/session/otp/asd1231dasd123"
expect(response.status).to eq(404)
post "/session/otp/asd1231dasd123"
expect(response.status).to eq(404)
end
context "when token is valid" do
it "should display the form for GET" do
token = SecureRandom.hex
Discourse.redis.setex "otp_#{token}", 10.minutes, user.username
get "/session/otp/#{token}"
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(response.body).to include(
I18n.t("user_api_key.otp_confirmation.logging_in_as", username: user.username),
)
expect(Discourse.redis.get("otp_#{token}")).to eq(user.username)
expect(session[:current_user_id]).to eq(nil)
end
it "should redirect on GET if already logged in" do
sign_in(user)
token = SecureRandom.hex
Discourse.redis.setex "otp_#{token}", 10.minutes, user.username
get "/session/otp/#{token}"
expect(response.status).to eq(302)
expect(Discourse.redis.get("otp_#{token}")).to eq(nil)
expect(session[:current_user_id]).to eq(user.id)
end
it "should authenticate user and delete token" do
user = Fabricate(:user)
get "/session/current.json"
expect(response.status).to eq(404)
token = SecureRandom.hex
Discourse.redis.setex "otp_#{token}", 10.minutes, user.username
post "/session/otp/#{token}"
expect(response.status).to eq(302)
expect(response).to redirect_to("/")
expect(Discourse.redis.get("otp_#{token}")).to eq(nil)
get "/session/current.json"
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
end
end
end
end
describe "#forgot_password" do
context "when hide_email_address_taken is set" do
before { SiteSetting.hide_email_address_taken = true }
it "denies for username" do
post "/session/forgot_password.json", params: { login: user.username }
expect(response.status).to eq(400)
expect(Jobs::CriticalUserEmail.jobs.size).to eq(0)
end
it "allows for username when staff" do
sign_in(Fabricate(:admin))
post "/session/forgot_password.json", params: { login: user.username }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(Jobs::CriticalUserEmail.jobs.size).to eq(1)
end
it "allows for email" do
post "/session/forgot_password.json", params: { login: user.email }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
expect(Jobs::CriticalUserEmail.jobs.size).to eq(1)
end
end
it "raises an error without a username parameter" do
post "/session/forgot_password.json"
expect(response.status).to eq(400)
end
it "should correctly screen ips" do
ScreenedIpAddress.create!(
ip_address: "100.0.0.1",
action_type: ScreenedIpAddress.actions[:block],
)
post "/session/forgot_password.json",
params: {
login: "made_up",
},
headers: {
"REMOTE_ADDR" => "100.0.0.1",
}
expect(response.parsed_body).to eq(
{ "errors" => [I18n.t("login.reset_not_allowed_from_ip_address")] },
)
end
it "should correctly rate limits" do
RateLimiter.enable
RateLimiter.clear_all!
user = Fabricate(:user)
3.times do
post "/session/forgot_password.json", params: { login: user.username }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
end
post "/session/forgot_password.json", params: { login: user.username }
expect(response.status).to eq(422)
3.times do
post "/session/forgot_password.json",
params: {
login: user.username,
},
headers: {
"REMOTE_ADDR" => "10.1.1.1",
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
end
post "/session/forgot_password.json",
params: {
login: user.username,
},
headers: {
"REMOTE_ADDR" => "100.1.1.1",
}
# not allowed, max 6 a day
expect(response.status).to eq(422)
end
context "for a non existant username" do
it "doesn't generate a new token for a made up username" do
expect do
post "/session/forgot_password.json", params: { login: "made_up" }
end.not_to change(EmailToken, :count)
expect(Jobs::CriticalUserEmail.jobs.size).to eq(0)
end
end
context "for an existing username" do
fab!(:user) { Fabricate(:user) }
context "when local login is disabled" do
before do
SiteSetting.enable_local_logins = false
post "/session/forgot_password.json", params: { login: user.username }
end
it_behaves_like "failed to continue local login"
end
context "when SSO is enabled" do
before do
FEATURE: Rename 'Discourse SSO' to DiscourseConnect (#11978) The 'Discourse SSO' protocol is being rebranded to DiscourseConnect. This should help to reduce confusion when 'SSO' is used in the generic sense. This commit aims to: - Rename `sso_` site settings. DiscourseConnect specific ones are prefixed `discourse_connect_`. Generic settings are prefixed `auth_` - Add (server-side-only) backwards compatibility for the old setting names, with deprecation notices - Copy `site_settings` database records to the new names - Rename relevant translation keys - Update relevant translations This commit does **not** aim to: - Rename any Ruby classes or methods. This might be done in a future commit - Change any URLs. This would break existing integrations - Make any changes to the protocol. This would break existing integrations - Change any functionality. Further normalization across DiscourseConnect and other auth methods will be done separately The risks are: - There is no backwards compatibility for site settings on the client-side. Accessing auth-related site settings in Javascript is fairly rare, and an error on the client side would not be security-critical. - If a plugin is monkey-patching parts of the auth process, changes to locale keys could cause broken error messages. This should also be unlikely. The old site setting names remain functional, so security-related overrides will remain working. A follow-up commit will be made with a post-deploy migration to delete the old `site_settings` rows.
2021-02-08 18:04:33 +08:00
SiteSetting.discourse_connect_url = "https://www.example.com/sso"
SiteSetting.enable_discourse_connect = true
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
end
it_behaves_like "failed to continue local login"
end
context "when local logins are disabled" do
before do
SiteSetting.enable_local_logins = false
post "/session.json", params: { login: user.username, password: "myawesomepassword" }
end
it_behaves_like "failed to continue local login"
end
context "when local logins via email are disabled" do
before { SiteSetting.enable_local_logins_via_email = false }
it "does not matter, generates a new token for a made up username" do
expect do
post "/session/forgot_password.json", params: { login: user.username }
end.to change(EmailToken, :count)
end
end
it "generates a new token for a made up username" do
expect do
post "/session/forgot_password.json", params: { login: user.username }
end.to change(EmailToken, :count)
end
it "enqueues an email" do
post "/session/forgot_password.json", params: { login: user.username }
expect(Jobs::CriticalUserEmail.jobs.size).to eq(1)
end
end
context "when doing nothing to system username" do
let(:system) { Discourse.system_user }
it "generates no token for system username" do
expect do
post "/session/forgot_password.json", params: { login: system.username }
end.not_to change(EmailToken, :count)
end
it "enqueues no email" do
post "/session/forgot_password.json", params: { login: system.username }
expect(Jobs::CriticalUserEmail.jobs.size).to eq(0)
end
end
context "for a staged account" do
let!(:staged) { Fabricate(:staged) }
it "generates no token for staged username" do
expect do
post "/session/forgot_password.json", params: { login: staged.username }
end.not_to change(EmailToken, :count)
end
it "enqueues no email" do
post "/session/forgot_password.json", params: { login: staged.username }
expect(Jobs::CriticalUserEmail.jobs.size).to eq(0)
end
end
end
describe "#current" do
context "when not logged in" do
it "returns 404" do
get "/session/current.json"
expect(response.status).to eq(404)
end
end
context "when logged in" do
let!(:user) { sign_in(Fabricate(:user)) }
it "returns the JSON for the user" do
get "/session/current.json"
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
json = response.parsed_body
expect(json["current_user"]).to be_present
expect(json["current_user"]["id"]).to eq(user.id)
end
end
end
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
describe "#second_factor_auth_show" do
let!(:user_second_factor) { Fabricate(:user_second_factor_totp, user: user) }
before { sign_in(user) }
it "returns 404 if there is no challenge for the given nonce" do
get "/session/2fa.json", params: { nonce: "asdasdsadsad" }
expect(response.status).to eq(404)
expect(response.parsed_body["error"]).to eq(I18n.t("second_factor_auth.challenge_not_found"))
end
it "returns 404 if the nonce does not match the challenge nonce" do
post "/session/2fa/test-action"
get "/session/2fa.json", params: { nonce: "wrongnonce" }
expect(response.status).to eq(404)
expect(response.parsed_body["error"]).to eq(I18n.t("second_factor_auth.challenge_not_found"))
end
it "returns 401 if the challenge nonce has expired" do
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
post "/session/2fa/test-action", xhr: true
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
nonce = response.parsed_body["second_factor_challenge_nonce"]
get "/session/2fa.json", params: { nonce: nonce }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
freeze_time (SecondFactor::AuthManager::MAX_CHALLENGE_AGE + 1.minute).from_now
get "/session/2fa.json", params: { nonce: nonce }
expect(response.status).to eq(401)
expect(response.parsed_body["error"]).to eq(I18n.t("second_factor_auth.challenge_expired"))
end
it "responds with challenge data" do
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
post "/session/2fa/test-action", xhr: true
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
nonce = response.parsed_body["second_factor_challenge_nonce"]
get "/session/2fa.json", params: { nonce: nonce }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
challenge_data = response.parsed_body
expect(challenge_data["totp_enabled"]).to eq(true)
expect(challenge_data["backup_enabled"]).to eq(false)
expect(challenge_data["security_keys_enabled"]).to eq(false)
expect(challenge_data["allowed_methods"]).to contain_exactly(
UserSecondFactor.methods[:totp],
UserSecondFactor.methods[:security_key],
)
expect(challenge_data["description"]).to eq("this is description for test action")
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
Fabricate(
:user_security_key_with_random_credential,
user: user,
name: "Enabled YubiKey",
enabled: true,
)
Fabricate(:user_second_factor_backup, user: user)
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
post "/session/2fa/test-action", params: { allow_backup_codes: true }, xhr: true
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
nonce = response.parsed_body["second_factor_challenge_nonce"]
get "/session/2fa.json", params: { nonce: nonce }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
challenge_data = response.parsed_body
expect(challenge_data["totp_enabled"]).to eq(true)
expect(challenge_data["backup_enabled"]).to eq(true)
expect(challenge_data["security_keys_enabled"]).to eq(true)
expect(challenge_data["allowed_credential_ids"]).to be_present
expect(challenge_data["challenge"]).to be_present
expect(challenge_data["allowed_methods"]).to contain_exactly(
UserSecondFactor.methods[:totp],
UserSecondFactor.methods[:security_key],
UserSecondFactor.methods[:backup_codes],
)
end
end
describe "#second_factor_auth_perform" do
let!(:user_second_factor) { Fabricate(:user_second_factor_totp, user: user) }
before { sign_in(user) }
it "returns 401 if the challenge nonce has expired" do
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
post "/session/2fa/test-action", xhr: true
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
nonce = response.parsed_body["second_factor_challenge_nonce"]
freeze_time (SecondFactor::AuthManager::MAX_CHALLENGE_AGE + 1.minute).from_now
token = ROTP::TOTP.new(user_second_factor.data).now
post "/session/2fa.json",
params: {
nonce: nonce,
second_factor_method: UserSecondFactor.methods[:totp],
second_factor_token: token,
}
expect(response.status).to eq(401)
expect(response.parsed_body["error"]).to eq(I18n.t("second_factor_auth.challenge_expired"))
end
it "returns 403 if the 2FA method is not allowed" do
Fabricate(:user_second_factor_backup, user: user)
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
post "/session/2fa/test-action", xhr: true
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
nonce = response.parsed_body["second_factor_challenge_nonce"]
post "/session/2fa.json",
params: {
nonce: nonce,
second_factor_method: UserSecondFactor.methods[:backup_codes],
second_factor_token: "iAmValidBackupCode",
}
expect(response.status).to eq(403)
end
it "returns 403 if the user disables the 2FA method in the middle of the 2FA process" do
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
post "/session/2fa/test-action", xhr: true
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
nonce = response.parsed_body["second_factor_challenge_nonce"]
token = ROTP::TOTP.new(user_second_factor.data).now
user_second_factor.destroy!
post "/session/2fa.json",
params: {
nonce: nonce,
second_factor_method: UserSecondFactor.methods[:totp],
second_factor_token: token,
}
expect(response.status).to eq(403)
end
it "marks the challenge as successful if the 2fa succeeds" do
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
post "/session/2fa/test-action", params: { redirect_url: "/ggg" }, xhr: true
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
nonce = response.parsed_body["second_factor_challenge_nonce"]
token = ROTP::TOTP.new(user_second_factor.data).now
post "/session/2fa.json",
params: {
nonce: nonce,
second_factor_method: UserSecondFactor.methods[:totp],
second_factor_token: token,
}
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
expect(response.parsed_body["ok"]).to eq(true)
expect(response.parsed_body["callback_method"]).to eq("POST")
expect(response.parsed_body["callback_path"]).to eq("/session/2fa/test-action")
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
expect(response.parsed_body["redirect_url"]).to eq("/ggg")
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
post "/session/2fa/test-action", params: { second_factor_nonce: nonce }
expect(response.status).to eq(200)
expect(response.parsed_body["error"]).not_to be_present
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
expect(response.parsed_body["result"]).to eq("second_factor_auth_completed")
end
it "does not mark the challenge as successful if the 2fa fails" do
FEATURE: Add 2FA support to the Discourse Connect Provider protocol (#16386) Discourse has the Discourse Connect Provider protocol that makes it possible to use a Discourse instance as an identity provider for external sites. As a natural extension to this protocol, this PR adds a new feature that makes it possible to use Discourse as a 2FA provider as well as an identity provider. The rationale for this change is that it's very difficult to implement 2FA support in a website and if you have multiple websites that need to have 2FA, it's unrealistic to build and maintain a separate 2FA implementation for each one. But with this change, you can piggyback on Discourse to take care of all the 2FA details for you for as many sites as you wish. To use Discourse as a 2FA provider, you'll need to follow this guide: https://meta.discourse.org/t/-/32974. It walks you through what you need to implement on your end/site and how to configure your Discourse instance. Once you're done, there is only one additional thing you need to do which is to include `require_2fa=true` in the payload that you send to Discourse. When Discourse sees `require_2fa=true`, it'll prompt the user to confirm their 2FA using whatever methods they've enabled (TOTP or security keys), and once they confirm they'll be redirected back to the return URL you've configured and the payload will contain `confirmed_2fa=true`. If the user has no 2FA methods enabled however, the payload will not contain `confirmed_2fa`, but it will contain `no_2fa_methods=true`. You'll need to be careful to re-run all the security checks and ensure the user can still access the resource on your site after they return from Discourse. This is very important because there's nothing that guarantees the user that will come back from Discourse after they confirm 2FA is the same user that you've redirected to Discourse. Internal ticket: t62183.
2022-04-13 20:04:09 +08:00
post "/session/2fa/test-action", params: { redirect_url: "/ggg" }, xhr: true
FEATURE: Centralized 2FA page (#15377) 2FA support in Discourse was added and grown gradually over the years: we first added support for TOTP for logins, then we implemented backup codes, and last but not least, security keys. 2FA usage was initially limited to logging in, but it has been expanded and we now require 2FA for risky actions such as adding a new admin to the site. As a result of this gradual growth of the 2FA system, technical debt has accumulated to the point where it has become difficult to require 2FA for more actions. We now have 5 different 2FA UI implementations and each one has to support all 3 2FA methods (TOTP, backup codes, and security keys) which makes it difficult to maintain a consistent UX for these different implementations. Moreover, there is a lot of repeated logic in the server-side code behind these 5 UI implementations which hinders maintainability even more. This commit is the first step towards repaying the technical debt: it builds a system that centralizes as much as possible of the 2FA server-side logic and UI. The 2 main components of this system are: 1. A dedicated page for 2FA with support for all 3 methods. 2. A reusable server-side class that centralizes the 2FA logic (the `SecondFactor::AuthManager` class). From a top-level view, the 2FA flow in this new system looks like this: 1. User initiates an action that requires 2FA; 2. Server is aware that 2FA is required for this action, so it redirects the user to the 2FA page if the user has a 2FA method, otherwise the action is performed. 3. User submits the 2FA form on the page; 4. Server validates the 2FA and if it's successful, the action is performed and the user is redirected to the previous page. A more technically-detailed explanation/documentation of the new system is available as a comment at the top of the `lib/second_factor/auth_manager.rb` file. Please note that the details are not set in stone and will likely change in the future, so please don't use the system in your plugins yet. Since this is a new system that needs to be tested, we've decided to migrate only the 2FA for adding a new admin to the new system at this time (in this commit). Our plan is to gradually migrate the remaining 2FA implementations to the new system. For screenshots of the 2FA page, see PR #15377 on GitHub.
2022-02-17 17:12:59 +08:00
nonce = response.parsed_body["second_factor_challenge_nonce"]
token = ROTP::TOTP.new(user_second_factor.data).now.to_i
token += token == 999_999 ? -1 : 1
post "/session/2fa.json",
params: {
nonce: nonce,
second_factor_method: UserSecondFactor.methods[:totp],
second_factor_token: token.to_s,
}
expect(response.status).to eq(400)
expect(response.parsed_body["ok"]).to eq(false)
expect(response.parsed_body["reason"]).to eq("invalid_second_factor")
expect(response.parsed_body["error"]).to eq(I18n.t("login.invalid_second_factor_code"))
post "/session/2fa/test-action", params: { second_factor_nonce: nonce }
expect(response.status).to eq(401)
end
end
describe "#scopes" do
context "when not a valid api request" do
it "returns 404" do
get "/session/scopes.json"
expect(response.status).to eq(404)
end
end
context "when a valid api request" do
let(:admin) { Fabricate(:admin) }
let(:scope) do
ApiKeyScope.new(resource: "topics", action: "read", allowed_parameters: { topic_id: "3" })
end
let(:api_key) { Fabricate(:api_key, user: admin, api_key_scopes: [scope]) }
it "returns the scopes of the api key" do
get "/session/scopes.json",
headers: {
"Api-Key": api_key.key,
"Api-Username": admin.username,
}
expect(response.status).to eq(200)
json = response.parsed_body
expect(json["scopes"].size).to eq(1)
expect(json["scopes"].first["resource"]).to eq("topics")
expect(json["scopes"].first["action"]).to eq("read")
expect(json["scopes"].first["allowed_parameters"]).to eq({ topic_id: "3" }.as_json)
end
end
end
end