perf: avoid unnecessary parsing of user-agents (#4449)

Author: sandhose

crates/data-model/src/compat/session.rs

@@ -11,7 +11,7 @@ use serde::Serialize;
 use ulid::Ulid;
 
 use super::Device;
-use crate::{InvalidTransitionError, UserAgent};
+use crate::InvalidTransitionError;
 
 #[derive(Debug, Clone, Default, PartialEq, Eq, Serialize)]
 pub enum CompatSessionState {
@@ -76,7 +76,7 @@ pub struct CompatSession {
     pub user_session_id: Option<Ulid>,
     pub created_at: DateTime<Utc>,
     pub is_synapse_admin: bool,
-    pub user_agent: Option<UserAgent>,
+    pub user_agent: Option<String>,
     pub last_active_at: Option<DateTime<Utc>>,
     pub last_active_ip: Option<IpAddr>,
 }

crates/data-model/src/oauth2/device_code_grant.rs

@@ -11,7 +11,7 @@ use oauth2_types::scope::Scope;
 use serde::Serialize;
 use ulid::Ulid;
 
-use crate::{BrowserSession, InvalidTransitionError, Session, UserAgent};
+use crate::{BrowserSession, InvalidTransitionError, Session};
 
 #[derive(Debug, Clone, PartialEq, Eq, Serialize)]
 #[serde(rename_all = "snake_case", tag = "state")]
@@ -192,7 +192,7 @@ pub struct DeviceCodeGrant {
     pub ip_address: Option<IpAddr>,
 
     /// The user agent used to request this device code grant.
-    pub user_agent: Option<UserAgent>,
+    pub user_agent: Option<String>,
 }
 
 impl std::ops::Deref for DeviceCodeGrant {

crates/data-model/src/oauth2/session.rs

@@ -11,7 +11,7 @@ use oauth2_types::scope::Scope;
 use serde::Serialize;
 use ulid::Ulid;
 
-use crate::{InvalidTransitionError, UserAgent};
+use crate::InvalidTransitionError;
 
 #[derive(Debug, Clone, Default, PartialEq, Eq, Serialize)]
 pub enum SessionState {
@@ -80,7 +80,7 @@ pub struct Session {
     pub user_session_id: Option<Ulid>,
     pub client_id: Ulid,
     pub scope: Scope,
-    pub user_agent: Option<UserAgent>,
+    pub user_agent: Option<String>,
     pub last_active_at: Option<DateTime<Utc>>,
     pub last_active_ip: Option<IpAddr>,
 }

crates/data-model/src/user_agent.rs

@@ -4,9 +4,18 @@
 // SPDX-License-Identifier: AGPL-3.0-only
 // Please see LICENSE in the repository root for full details.
 
+use std::sync::LazyLock;
+
 use serde::Serialize;
 use woothee::{parser::Parser, woothee::VALUE_UNKNOWN};
 
+static CUSTOM_USER_AGENT_REGEX: LazyLock<regex::Regex> = LazyLock::new(|| {
+    regex::Regex::new(r"^(?P<name>[^/]+)/(?P<version>[^ ]+) \((?P<segments>.+)\)$").unwrap()
+});
+
+static ELECTRON_USER_AGENT_REGEX: LazyLock<regex::Regex> =
+    LazyLock::new(|| regex::Regex::new(r"(?m)\w+/[\w.]+").unwrap());
+
 #[derive(Debug, Serialize, Clone, PartialEq, Eq)]
 #[serde(rename_all = "snake_case")]
 pub enum DeviceType {
@@ -37,10 +46,7 @@ impl std::ops::Deref for UserAgent {
 
 impl UserAgent {
     fn parse_custom(user_agent: &str) -> Option<(&str, &str, &str, &str, Option<&str>)> {
-        let regex = regex::Regex::new(r"^(?P<name>[^/]+)/(?P<version>[^ ]+) \((?P<segments>.+)\)$")
-            .unwrap();
-
-        let captures = regex.captures(user_agent)?;
+        let captures = CUSTOM_USER_AGENT_REGEX.captures(user_agent)?;
         let name = captures.name("name")?.as_str();
         let version = captures.name("version")?.as_str();
         let segments: Vec<&str> = captures
@@ -73,9 +79,8 @@ impl UserAgent {
     }
 
     fn parse_electron(user_agent: &str) -> Option<(&str, &str)> {
-        let regex = regex::Regex::new(r"(?m)\w+/[\w.]+").unwrap();
         let omit_keys = ["Mozilla", "AppleWebKit", "Chrome", "Electron", "Safari"];
-        return regex
+        return ELECTRON_USER_AGENT_REGEX
             .find_iter(user_agent)
             .map(|caps| caps.as_str().split_once('/').unwrap())
             .find(|pair| !omit_keys.contains(&pair.0));

crates/data-model/src/users.rs

@@ -12,8 +12,6 @@ use serde::Serialize;
 use ulid::Ulid;
 use url::Url;
 
-use crate::UserAgent;
-
 #[derive(Debug, Clone, PartialEq, Eq, Serialize)]
 pub struct User {
     pub id: Ulid,
@@ -81,7 +79,7 @@ pub enum AuthenticationMethod {
 pub struct UserRecoverySession {
     pub id: Ulid,
     pub email: String,
-    pub user_agent: UserAgent,
+    pub user_agent: String,
     pub ip_address: Option<IpAddr>,
     pub locale: String,
     pub created_at: DateTime<Utc>,
@@ -137,7 +135,7 @@ pub struct BrowserSession {
     pub user: User,
     pub created_at: DateTime<Utc>,
     pub finished_at: Option<DateTime<Utc>>,
-    pub user_agent: Option<UserAgent>,
+    pub user_agent: Option<String>,
     pub last_active_at: Option<DateTime<Utc>>,
     pub last_active_ip: Option<IpAddr>,
 }
@@ -159,9 +157,9 @@ impl BrowserSession {
                 user,
                 created_at: now,
                 finished_at: None,
-                user_agent: Some(UserAgent::parse(
+                user_agent: Some(
                     "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/93.0.0.0 Safari/537.36".to_owned()
-                )),
+                ),
                 last_active_at: Some(now),
                 last_active_ip: None,
             })
@@ -213,7 +211,7 @@ pub struct UserRegistration {
     pub password: Option<UserRegistrationPassword>,
     pub post_auth_action: Option<serde_json::Value>,
     pub ip_address: Option<IpAddr>,
-    pub user_agent: Option<UserAgent>,
+    pub user_agent: Option<String>,
     pub created_at: DateTime<Utc>,
     pub completed_at: Option<DateTime<Utc>>,
 }

crates/handlers/src/admin/model.rs

@@ -206,7 +206,7 @@ impl
             user_session_id: session.user_session_id,
             redirect_uri: sso_login.map(|sso| sso.redirect_uri),
             created_at: session.created_at,
-            user_agent: session.user_agent.map(|ua| ua.raw),
+            user_agent: session.user_agent,
             last_active_at: session.last_active_at,
             last_active_ip: session.last_active_ip,
             finished_at,
@@ -313,7 +313,7 @@ impl From<mas_data_model::Session> for OAuth2Session {
             user_session_id: session.user_session_id,
             client_id: session.client_id,
             scope: session.scope.to_string(),
-            user_agent: session.user_agent.map(|ua| ua.raw),
+            user_agent: session.user_agent,
             last_active_at: session.last_active_at,
             last_active_ip: session.last_active_ip,
         }
@@ -406,7 +406,7 @@ impl From<mas_data_model::BrowserSession> for UserSession {
             created_at: value.created_at,
             finished_at: value.finished_at,
             user_id: value.user.id,
-            user_agent: value.user_agent.map(|ua| ua.raw),
+            user_agent: value.user_agent,
             last_active_at: value.last_active_at,
             last_active_ip: value.last_active_ip,
         }

crates/handlers/src/compat/login.rs

@@ -11,9 +11,7 @@ use axum_extra::typed_header::TypedHeader;
 use chrono::Duration;
 use hyper::StatusCode;
 use mas_axum_utils::record_error;
-use mas_data_model::{
-    CompatSession, CompatSsoLoginState, Device, SiteConfig, TokenType, User, UserAgent,
-};
+use mas_data_model::{CompatSession, CompatSsoLoginState, Device, SiteConfig, TokenType, User};
 use mas_matrix::HomeserverConnection;
 use mas_storage::{
     BoxClock, BoxRepository, BoxRng, Clock, RepositoryAccess,
@@ -276,7 +274,7 @@ pub(crate) async fn post(
     user_agent: Option<TypedHeader<headers::UserAgent>>,
     MatrixJsonBody(input): MatrixJsonBody<RequestBody>,
 ) -> Result<impl IntoResponse, RouteError> {
-    let user_agent = user_agent.map(|ua| UserAgent::parse(ua.as_str().to_owned()));
+    let user_agent = user_agent.map(|ua| ua.as_str().to_owned());
     let login_type = input.credentials.login_type();
     let (mut session, user) = match (password_manager.is_enabled(), input.credentials) {
         (

crates/handlers/src/graphql/model/browser_sessions.rs

@@ -81,7 +81,11 @@ impl BrowserSession {
 
     /// The user-agent with which the session was created.
     pub async fn user_agent(&self) -> Option<UserAgent> {
-        self.0.user_agent.clone().map(UserAgent::from)
+        self.0
+            .user_agent
+            .clone()
+            .map(mas_data_model::UserAgent::parse)
+            .map(UserAgent::from)
     }
 
     /// The last IP address used by the session.

crates/handlers/src/graphql/model/compat_sessions.rs

@@ -98,7 +98,11 @@ impl CompatSession {
 
     /// The user-agent with which the session was created.
     pub async fn user_agent(&self) -> Option<UserAgent> {
-        self.session.user_agent.clone().map(UserAgent::from)
+        self.session
+            .user_agent
+            .clone()
+            .map(mas_data_model::UserAgent::parse)
+            .map(UserAgent::from)
     }
 
     /// The associated SSO login, if any.

crates/handlers/src/graphql/model/oauth.rs

@@ -61,7 +61,11 @@ impl OAuth2Session {
 
     /// The user-agent with which the session was created.
     pub async fn user_agent(&self) -> Option<UserAgent> {
-        self.0.user_agent.clone().map(UserAgent::from)
+        self.0
+            .user_agent
+            .clone()
+            .map(mas_data_model::UserAgent::parse)
+            .map(UserAgent::from)
     }
 
     /// The state of the session.