0.7.0: Ununsafe urouter

Signed-off-by: Ivan Bushchik <ivabus@ivabus.dev>

.gitignore

@@ -1 +1,3 @@
 /target
+.idea/
+*.DS_Store

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "urouter"
-version = "0.5.3"
+version = "0.7.0"
 edition = "2021"
 license = "MIT"
 repository = "https://github.com/ivabus/urouter"
@@ -9,10 +9,11 @@ description = "Small HTTP router"
 
 [dependencies]
 rocket = "0.5.0"
-serde = { version = "1.0.193", features = ["derive"] }
-serde_json = "1.0.108"
+serde = { version = "1.0.197", features = ["derive"] }
+serde_json = "1.0.114"
 url-escape = "0.1.1"
-smurf = "0.3.0"
-clap = { version = "4.4.11", features = ["derive"] }
-regex = "1.10.2"
-ureq = { version = "2.9.1", features = ["brotli", "native-certs"] }
+once_cell = "1.19"
+clap = { version = "4.5.3", features = ["derive"] }
+regex = "1.10.3"
+ureq = { version = "2.9.6", features = ["brotli", "native-certs"] }
+users = "0.11.0"

README.md

@@ -21,6 +21,7 @@ Each set contains 2 necessary elements and 1 optional.
   - `url` (string) - redirect to URL with HTTP 303 See Other
   - `file` (string) - read file from path `--dir/file` where `--dir` is option (default: `.`, see `--help`) and respond with HTTP 200 OK with `content-type: text/plain`
   - `text` (string) - plain text with HTTP 200 OK with `content-type: text/plain`
+  - `html` (string) - plain text with HTTP 200 OK with `content-type: text/html`
   - `external` (set) - download (every time) file using `ureq` HTTP library and response with contents of downloaded resource with HTTP 200 OK and extracted `content-type` from response
     - `url` (string) - URL to download
     - `headers` (set, optional) - headers to include with request
@@ -85,6 +86,14 @@ Each set contains 2 necessary elements and 1 optional.
 
 Agent matching made for `curl https://url | sh` like scripts.
 
+## `alias.json` location
+
+- Passed with `--alias_file`, will look up to this path, if file doesn't exist (or no access to it) will panic
+- If urouter started with privileges (EUID = 0), file would be `/etc/urouter/alias.json`
+- Otherwise if `XDG_CONFIG_HOME` is set, file would be `$XDG_CONFIG_HOME/urouter/alias.json`
+- Otherwise if `$HOME` is set, file would be `$HOME/.config/urouter/alias.json`
+- If not matched any, will panic and exit
+
 ## License
 
 The project is licensed under the terms of the [MIT license](./LICENSE).

alias.json

@@ -17,7 +17,7 @@
   {
     "uri":"text",
     "alias": {
-      "text": "sometext"
+      "html": "sometext"
     },
     "agent": {
       "regex": "^curl/[0-9].[0-9].[0-9]$",

src/main.rs

@@ -6,28 +6,62 @@ Global comments:
   500 Internal Server Error
 */
 
-mod structs;
+#![forbid(unsafe_code)]
 
+mod structs;
 use structs::*;
+
 #[macro_use]
 extern crate rocket;
 
-use rocket::http::{ContentType, Status};
-use std::cell::OnceCell;
-use std::collections::HashMap;
-use std::hint::unreachable_unchecked;
-use std::path::PathBuf;
-use std::time::Instant;
+use std::io::Write;
+use std::sync::Arc;
+use std::{collections::HashMap, path::PathBuf, time::Instant};
 
-use rocket::response::content::RawText;
-use rocket::response::Redirect;
+use rocket::http::{ContentType, Status};
+use rocket::{
+	figment::Figment,
+	response::{content::RawHtml, content::RawText, Redirect},
+};
 
 use clap::Parser;
+use once_cell::sync::Lazy;
 use regex::Regex;
-use rocket::figment::Figment;
 
-static mut ALIAS: OnceCell<Vec<Alias>> = OnceCell::new();
-static mut COMPILED_REGEXES: OnceCell<HashMap<String, Regex>> = OnceCell::new();
+static ALIAS: Lazy<Arc<Vec<Alias>>> = Lazy::new(|| {
+	let mut args = Args::parse();
+	if args.alias_file.is_none() {
+		args.alias_file = Some(get_config_file_location());
+	}
+
+	let file = std::fs::File::open(args.alias_file.unwrap()).unwrap();
+	let alias: Vec<Alias> = if args.alias_file_is_set_not_a_list {
+		serde_json::from_reader::<std::fs::File, NixJson>(file).unwrap().alias
+	} else {
+		serde_json::from_reader::<std::fs::File, Vec<Alias>>(file).unwrap()
+	};
+	Arc::new(alias)
+});
+static COMPILED_REGEXES: Lazy<Arc<HashMap<String, Regex>>> = Lazy::new(|| {
+	let compilation_start = Instant::now();
+	let mut regexes_len = 0;
+	// Precompile all regexes
+	let mut compiled_regexes: HashMap<String, Regex> = HashMap::new();
+	for i in &**ALIAS {
+		if let Some(agent) = &i.agent {
+			compiled_regexes.insert(agent.regex.clone(), Regex::new(&agent.regex).unwrap());
+			regexes_len += 1;
+		}
+	}
+	if regexes_len != 0 {
+		println!(
+			"Compiled {} regexes in {} ms",
+			regexes_len,
+			(Instant::now() - compilation_start).as_secs_f64() * 1000.0
+		);
+	}
+	Arc::new(compiled_regexes)
+});
 
 fn get_return(alias: &Alias) -> Response {
 	let args = Args::parse();
@@ -36,9 +70,10 @@ fn get_return(alias: &Alias) -> Response {
 		AliasType::Url(url) => Response::Redirect(Box::from(Redirect::to(url.clone()))),
 		AliasType::File(path) => {
 			dir.push(&PathBuf::from(&path));
-			Response::Text(Box::new(RawText(smurf::io::read_file_str(&dir).unwrap())))
+			Response::Text(Box::new(RawText(std::fs::read_to_string(&dir).unwrap())))
 		}
 		AliasType::Text(text) => Response::Text(Box::new(RawText(text.clone()))),
+		AliasType::Html(html) => Response::Html(Box::new(RawHtml(html.clone()))),
 		AliasType::External(source) => {
 			let mut request = ureq::get(&source.url);
 			for (header, value) in &source.headers {
@@ -58,29 +93,22 @@ fn get_return(alias: &Alias) -> Response {
 fn get_page(page: &str, user_agent: UserAgent) -> Response {
 	let mut decoded_page = String::new();
 	url_escape::decode_to_string(page, &mut decoded_page);
-	let alias = unsafe { ALIAS.get().unwrap() };
 	let mut pages = Vec::new();
-	for i in alias {
+	for i in &**ALIAS {
 		if i.uri == decoded_page {
 			if let Some(agent) = &i.agent {
-				unsafe {
-					let regexes = COMPILED_REGEXES.get_mut();
-					let re = if let Some(r) = regexes {
-						// Unwrapping safely, guaranteed to be generated during initialization
-						r.get(&agent.regex).unwrap()
-					} else {
-						unreachable_unchecked()
-					};
+				let regexes = &COMPILED_REGEXES;
+				let re = regexes.get(&agent.regex).unwrap();
 
-					if re.is_match(&user_agent.0) {
-						return get_return(i);
-					}
+				if re.is_match(&user_agent.0) {
+					return get_return(i);
+				}
 
-					if let Some(true) = agent.only_matching {
-						continue;
-					}
+				if let Some(true) = agent.only_matching {
+					continue;
 				}
 			}
+
 			pages.push(i);
 		}
 	}
@@ -96,37 +124,31 @@ async fn index(user_agent: UserAgent) -> Response {
 	get_page("/", user_agent)
 }
 
+fn get_config_file_location() -> PathBuf {
+	if users::get_effective_uid() == 0 {
+		return "/etc/urouter/alias.json".parse().unwrap();
+	}
+
+	if let Ok(config_home) = std::env::var("XDG_CONFIG_HOME") {
+		return format!("{}/urouter/alias.json", config_home).parse().unwrap();
+	}
+
+	if let Ok(home) = std::env::var("HOME") {
+		return format!("{}/.config/urouter/alias.json", home).parse().unwrap();
+	}
+
+	if !std::path::Path::new("alias.json").exists() {
+		let mut file = std::fs::File::create("alias.json").unwrap();
+		file.write_all(b"[]").unwrap();
+	}
+	"alias.json".parse().unwrap()
+}
+
 #[rocket::main]
 async fn main() -> Result<(), rocket::Error> {
 	let args = Args::parse();
-	let file = std::fs::File::open(args.alias_file).unwrap();
-	let alias: Vec<Alias> = if args.alias_file_is_set_not_a_list {
-		serde_json::from_reader::<std::fs::File, NixJson>(file).unwrap().alias
-	} else {
-		serde_json::from_reader::<std::fs::File, Vec<Alias>>(file).unwrap()
-	};
-	unsafe {
-		ALIAS.set(alias).unwrap();
-
-		let compilation_start = Instant::now();
-		let mut regexes_len = 0;
-		// Precompile all regexes
-		let mut compiled_regexes: HashMap<String, Regex> = HashMap::new();
-		for i in ALIAS.get().unwrap() {
-			if let Some(agent) = &i.agent {
-				compiled_regexes.insert(agent.regex.clone(), Regex::new(&agent.regex).unwrap());
-				regexes_len += 1;
-			}
-		}
-		if regexes_len != 0 {
-			println!(
-				"Compiled {} regexes in {} ms",
-				regexes_len,
-				(Instant::now() - compilation_start).as_secs_f64() * 1000.0
-			);
-		}
-		COMPILED_REGEXES.set(compiled_regexes).unwrap();
-	}
+	let _alias = &**ALIAS;
+	let _regex = &**COMPILED_REGEXES;
 
 	let figment = Figment::from(rocket::Config::default())
 		.merge(("ident", format!("urouter/{}", env!("CARGO_PKG_VERSION"))))

src/structs.rs

@@ -3,7 +3,7 @@
 use clap::Parser;
 use rocket::http::{ContentType, Status};
 use rocket::request::{FromRequest, Outcome};
-use rocket::response::content::RawText;
+use rocket::response::content::{RawHtml, RawText};
 use rocket::response::{Redirect, Responder};
 use rocket::Request;
 use serde::Deserialize;
@@ -14,8 +14,8 @@ use std::path::PathBuf;
 #[derive(Parser, Debug)]
 #[command(about, author)]
 pub struct Args {
-	#[arg(long, default_value = "./alias.json")]
-	pub alias_file: PathBuf,
+	#[arg(long)]
+	pub alias_file: Option<PathBuf>,
 
 	/// For internal usage
 	#[arg(long, default_value = "false")]
@@ -53,6 +53,8 @@ pub enum AliasType {
 	File(String),
 	#[serde(alias = "text")]
 	Text(String),
+	#[serde(alias = "html")]
+	Html(String),
 	#[serde(alias = "external")]
 	External(External),
 }
@@ -73,6 +75,7 @@ pub struct External {
 #[derive(Responder)]
 pub enum Response {
 	Text(Box<RawText<String>>),
+	Html(Box<RawHtml<String>>),
 	Redirect(Box<Redirect>),
 	Status(Status),
 	Custom(Box<(ContentType, RawText<String>)>),