Antispam

README.md

@@ -11,6 +11,7 @@ Rust webserver for comments, that you can easily embed in a website.
 * Admin notification on new comment via Matrix
 * Embedded one-file webserver
 * [Tera](https://github.com/Keats/tera) templates
+* Comment frequency limit per IP
 
 ## Use
 

src/cleaner.rs

@@ -0,0 +1,30 @@
+use crate::{config::Config, db::*};
+
+use std::{sync::Arc, time::Duration};
+
+pub async fn run_cleaner(config: Arc<Config>, dbs: Dbs) {
+	let mut last_db_clean = 0;
+	loop {
+		let time = std::time::SystemTime::now()
+			.duration_since(std::time::UNIX_EPOCH)
+			.unwrap()
+			.as_secs();
+
+		if time > last_db_clean + 3600 {
+			clean_antispam(config.clone(), dbs.clone(), time);
+			last_db_clean = time;
+		}
+
+		tokio::time::sleep(Duration::from_secs(60)).await;
+	}
+}
+
+fn clean_antispam(config: Arc<Config>, dbs: Dbs, time: u64) {
+	for (addr, (last_mutation, _mutation_count)) in
+		dbs.client_mutation.iter().filter_map(|o| o.ok())
+	{
+		if last_mutation + config.antispam_duration < time {
+			dbs.client_mutation.remove(&addr).unwrap();
+		}
+	}
+}

src/config.rs

@@ -8,10 +8,18 @@ const CONFIG_FILE: &str = "config.toml";
 
 #[derive(Deserialize, Serialize)]
 pub struct Config {
-	//#[serde(default = "Config::default_admin_emails")]
-	//pub admin_emails: Vec<String>,
 	#[serde(default = "Config::default_admin_passwords")]
 	pub admin_passwords: Vec<String>,
+	/// (seconds)
+	#[serde(default = "Config::default_antispam_duration")]
+	pub antispam_duration: u64,
+	#[serde(default = "Config::default_antispam_enable")]
+	pub antispam_enable: bool,
+	/// Maximum number of mutations by IP within antispam_duration
+	#[serde(default = "Config::default_antispam_mutation_limit")]
+	pub antispam_mutation_limit: u32,
+	#[serde(default = "Config::default_antispam_whitelist")]
+	pub antispam_whitelist: Vec<IpAddr>,
 	/// New or edited comments need admin's approval before being public
 	#[serde(default = "Config::default_comment_approve")]
 	pub comment_approve: bool,
@@ -43,17 +51,30 @@ pub struct Config {
 	pub matrix_server: String,
 	#[serde(default = "Config::default_matrix_user")]
 	pub matrix_user: String,
+	/// Are we behind a reverse proxy?
+	/// Determines whether we assume client address is in a Forwarded header or socket address.
+	#[serde(default = "Config::default_reverse_proxy")]
+	pub reverse_proxy: bool,
 	#[serde(default = "Config::default_root_url")]
 	pub root_url: String,
 }
 
 impl Config {
-	/*fn default_admin_emails() -> Vec<String> {
-		vec![]
-	}*/
 	fn default_admin_passwords() -> Vec<String> {
 		vec![]
 	}
+	fn default_antispam_duration() -> u64 {
+		3600
+	}
+	fn default_antispam_enable() -> bool {
+		true
+	}
+	fn default_antispam_mutation_limit() -> u32 {
+		10
+	}
+	fn default_antispam_whitelist() -> Vec<IpAddr> {
+		vec![[127u8, 0, 0, 1].into(), [0u8; 4].into(), [0u8; 16].into()]
+	}
 	fn default_comment_approve() -> bool {
 		true
 	}
@@ -96,6 +117,9 @@ impl Config {
 	fn default_matrix_user() -> String {
 		"@tuxmain:matrix.txmn.tk".into()
 	}
+	fn default_reverse_proxy() -> bool {
+		false
+	}
 	fn default_root_url() -> String {
 		"/".into()
 	}
@@ -104,8 +128,11 @@ impl Config {
 impl Default for Config {
 	fn default() -> Self {
 		Self {
-			//admin_emails: Self::default_admin_emails(),
 			admin_passwords: Self::default_admin_passwords(),
+			antispam_duration: Self::default_antispam_duration(),
+			antispam_enable: Self::default_antispam_enable(),
+			antispam_mutation_limit: Self::default_antispam_mutation_limit(),
+			antispam_whitelist: Self::default_antispam_whitelist(),
 			comment_approve: Self::default_comment_approve(),
 			comment_edit_timeout: Self::default_comment_edit_timeout(),
 			comment_author_max_len: Self::default_comment_author_max_len(),
@@ -120,6 +147,7 @@ impl Default for Config {
 			matrix_room: Self::default_matrix_room(),
 			matrix_server: Self::default_matrix_server(),
 			matrix_user: Self::default_matrix_user(),
+			reverse_proxy: Self::default_reverse_proxy(),
 			root_url: Self::default_root_url(),
 		}
 	}

src/db.rs

@@ -1,6 +1,6 @@
 use serde::{Deserialize, Serialize};
 use sha2::{Digest, Sha256};
-use std::path::Path;
+use std::{net::IpAddr, path::Path};
 pub use typed_sled::Tree;
 
 const DB_DIR: &str = "db";
@@ -12,6 +12,8 @@ pub struct Dbs {
 	pub comment: Tree<CommentId, Comment>,
 	pub comment_approved: Tree<(TopicHash, Time, CommentId), ()>,
 	pub comment_pending: Tree<(TopicHash, Time, CommentId), ()>,
+	/// client_addr -> (last_mutation, mutation_count)
+	pub client_mutation: Tree<IpAddr, (Time, u32)>,
 }
 
 pub fn load_dbs(path: Option<&Path>) -> Dbs {
@@ -28,6 +30,7 @@ pub fn load_dbs(path: Option<&Path>) -> Dbs {
 		comment: Tree::open(&db, "comment"),
 		comment_approved: Tree::open(&db, "comment_approved"),
 		comment_pending: Tree::open(&db, "comment_pending"),
+		client_mutation: Tree::open(&db, "client_mutation"),
 	}
 }
 

src/helpers.rs

@@ -1,6 +1,7 @@
-use crate::db::*;
+use crate::{config::Config, db::*};
 
 use log::error;
+use std::{net::IpAddr, str::FromStr};
 
 pub fn new_pending_comment(comment: &Comment, dbs: &Dbs) -> Result<CommentId, sled::Error> {
 	let comment_id = CommentId::new();
@@ -80,7 +81,87 @@ pub fn iter_pending_comments_by_topic(
 	iter_comments_by_topic(topic_hash, &dbs.comment_pending, dbs)
 }
 
-//pub enum DbHelperError {}
+/// Returns Some(time_left) if the client is banned.
+pub fn antispam_check_client_mutation(
+	addr: &IpAddr,
+	dbs: &Dbs,
+	config: &Config,
+) -> Result<Option<Time>, sled::Error> {
+	let time = std::time::SystemTime::now()
+		.duration_since(std::time::UNIX_EPOCH)
+		.unwrap()
+		.as_secs();
+	Ok(dbs
+		.client_mutation
+		.get(addr)?
+		.and_then(|(last_mutation, mutation_count)| {
+			let timeout = last_mutation + config.antispam_duration;
+			if timeout > time && mutation_count >= config.antispam_mutation_limit {
+				Some(timeout - time)
+			} else {
+				None
+			}
+		}))
+}
+
+pub fn antispam_update_client_mutation(addr: &IpAddr, dbs: &Dbs) -> Result<(), sled::Error> {
+	let time = std::time::SystemTime::now()
+		.duration_since(std::time::UNIX_EPOCH)
+		.unwrap()
+		.as_secs();
+	dbs.client_mutation.fetch_and_update(addr, |entry| {
+		if let Some((_last_mutation, mutation_count)) = entry {
+			Some((time, mutation_count.saturating_add(1)))
+		} else {
+			Some((time, 1))
+		}
+	})?;
+	Ok(())
+}
+
+/*pub fn new_client_mutation(
+	addr: &IpAddr,
+	dbs: &Dbs,
+	config: &Config,
+) -> Result<Option<Time>, sled::Error> {
+	let time = std::time::SystemTime::now()
+		.duration_since(std::time::UNIX_EPOCH)
+		.unwrap()
+		.as_secs();
+	let mut res = None;
+	dbs.client_mutation.fetch_and_update(addr, |entry| {
+		if let Some((last_mutation, mutation_count)) = entry {
+			if last_mutation + config.antispam_duration > time {
+				if mutation_count >= config.antispam_mutation_limit {
+					res = Some(last_mutation + config.antispam_duration);
+					Some((last_mutation, mutation_count))
+				} else {
+					Some((time, mutation_count.saturating_add(1)))
+				}
+			} else {
+				Some((time, 1))
+			}
+		} else {
+			Some((time, 1))
+		}
+	})?;
+	Ok(res)
+}*/
+
+pub fn get_client_addr<State>(
+	config: &Config,
+	req: &tide::Request<State>,
+) -> Option<Result<IpAddr, std::net::AddrParseError>> {
+	Some(IpAddr::from_str(
+		if config.reverse_proxy {
+			req.remote()
+		} else {
+			req.peer_addr()
+		}?
+		.rsplit_once(':')?
+		.0,
+	))
+}
 
 #[cfg(test)]
 mod test {

src/main.rs

@@ -1,3 +1,4 @@
+mod cleaner;
 mod cli;
 mod config;
 mod db;
@@ -12,6 +13,7 @@ use argon2::{
 	Argon2,
 };
 use clap::Parser;
+use std::sync::Arc;
 
 #[tokio::main]
 async fn main() {
@@ -23,7 +25,10 @@ async fn main() {
 		}
 		cli::MainSubcommand::Start(subopt) => {
 			let (config, dbs, templates) = init_all(opt.opt, subopt);
-			server::start_server(config, dbs, templates).await
+			let config = Arc::new(config);
+			let templates = Arc::new(templates);
+			tokio::spawn(cleaner::run_cleaner(config.clone(), dbs.clone()));
+			server::run_server(config, dbs, templates).await;
 		}
 		cli::MainSubcommand::Psw => {
 			let mut config = config::read_config(&opt.opt.dir.0);

src/server.rs

@@ -2,16 +2,13 @@ use crate::{config::*, db::*, helpers, queries::*, templates::*};
 
 use argon2::{Argon2, PasswordHash, PasswordVerifier};
 use crossbeam_channel::Sender;
-use log::error;
+use log::{error, warn};
 use std::sync::Arc;
 use tera::Context;
 
-pub async fn start_server(config: Config, dbs: Dbs, templates: Templates) {
+pub async fn run_server(config: Arc<Config>, dbs: Dbs, templates: Arc<Templates>) {
 	tide::log::start();
 
-	let templates = Arc::new(templates);
-	let config = Arc::new(config);
-
 	let (notify_send, notify_recv) = crossbeam_channel::bounded(10);
 	tokio::spawn(crate::notify::run_notifier(config.clone(), notify_recv));
 
@@ -61,6 +58,8 @@ async fn serve_comments<'a>(
 	dbs: Dbs,
 	errors: &[String],
 ) -> tide::Result<tide::Response> {
+	dbg!(req.peer_addr());
+
 	let Ok(topic) = req.param("topic") else {
 		return Err(tide::Error::from_str(404, "No topic"))
 	};
@@ -119,10 +118,12 @@ async fn serve_comments<'a>(
 			.collect::<Vec<CommentWithId>>(),
 	);
 
-	Ok(tide::Response::builder(200)
-		.content_type(tide::http::mime::HTML)
-		.body(templates.tera.render("comments.html", &context)?)
-		.build())
+	Ok(
+		tide::Response::builder(if errors.is_empty() { 200 } else { 400 })
+			.content_type(tide::http::mime::HTML)
+			.body(templates.tera.render("comments.html", &context)?)
+			.build(),
+	)
 }
 
 async fn serve_admin<'a>(
@@ -190,6 +191,28 @@ async fn handle_post_comments(
 	dbs: Dbs,
 	notify_send: Sender<()>,
 ) -> tide::Result<tide::Response> {
+	let client_addr = if config.antispam_enable {
+		match helpers::get_client_addr(&config, &req) {
+			Some(Ok(addr)) => {
+				if config.antispam_whitelist.contains(&addr) {
+					None
+				} else {
+					Some(addr)
+				}
+			}
+			Some(Err(e)) => {
+				warn!("Unable to parse client addr: {}", e);
+				None
+			}
+			None => {
+				warn!("No client addr");
+				None
+			}
+		}
+	} else {
+		None
+	};
+
 	let mut errors = Vec::new();
 
 	match req.body_form::<CommentQuery>().await? {
@@ -219,8 +242,22 @@ async fn handle_post_comments(
 					config.comment_text_max_len
 				));
 			}
+			if let Some(client_addr) = &client_addr {
+				if let Some(antispam_timeout) =
+					helpers::antispam_check_client_mutation(client_addr, &dbs, &config).unwrap()
+				{
+					errors.push(format!(
+						"The edition quota from your IP is reached. You will be unblocked in {}s.",
+						antispam_timeout
+					));
+				}
+			}
 
 			if errors.is_empty() {
+				if let Some(client_addr) = &client_addr {
+					helpers::antispam_update_client_mutation(client_addr, &dbs).unwrap();
+				}
+
 				let topic_hash = TopicHash::from_topic(topic);
 
 				let time = std::time::SystemTime::now()