Files
RustProxy/src/transform/compile.rs
T
2026-10-05 09:30:21 +02:00

334 lines
10 KiB
Rust

pub trait Transformer {
fn transform(&self, body: &[u8]) -> Result<Vec<u8>>;
}
struct JsTransformer;
impl Transformer for JsTransformer {
fn transform(&self, body: &[u8]) -> Result<Vec<u8>> {
js::minify(body)
}
}
struct CssTransformer;
impl Transformer for CssTransformer {
fn transform(&self, body: &[u8]) -> Result<Vec<u8>> {
css::minify(body)
}
}
struct HtmlTransformer;
impl Transformer for HtmlTransformer {
fn transform(&self, body: &[u8]) -> Result<Vec<u8>> {
html::minify(body)
}
}
struct SvgTransformer;
impl Transformer for SvgTransformer {
fn transform(&self, body: &[u8]) -> Result<Vec<u8>> {
svg::optimize(body)
}
}
struct JsonTransformer;
impl Transformer for JsonTransformer {
fn transform(&self, body: &[u8]) -> Result<Vec<u8>> {
json::compact(body)
}
}
fn transformer(kind: Kind) -> &'static dyn Transformer {
static JS: JsTransformer = JsTransformer;
static CSS: CssTransformer = CssTransformer;
static HTML: HtmlTransformer = HtmlTransformer;
static SVG: SvgTransformer = SvgTransformer;
static JSON: JsonTransformer = JsonTransformer;
match kind {
Kind::Javascript => &JS,
Kind::Css => &CSS,
Kind::Html => &HTML,
Kind::Svg => &SVG,
Kind::Json => &JSON,
}
}
pub fn no_transform(headers: &HeaderMap) -> bool {
headers
.get_all("cache-control")
.iter()
.filter_map(|value| value.to_str().ok())
.any(|value| {
value.split(',').any(|directive| {
directive
.trim()
.split('=')
.next()
.is_some_and(|name| name.trim().eq_ignore_ascii_case("no-transform"))
})
})
}
fn content_type_kind(headers: &HeaderMap) -> Option<Kind> {
let value = headers.get("content-type")?.to_str().ok()?;
let mime = value.split(';').next()?.trim().to_ascii_lowercase();
match mime.as_str() {
"application/javascript"
| "text/javascript"
| "application/ecmascript"
| "text/ecmascript" => Some(Kind::Javascript),
"text/css" => Some(Kind::Css),
"text/html" | "application/xhtml+xml" => Some(Kind::Html),
"image/svg+xml" => Some(Kind::Svg),
"application/json" | "text/json" => Some(Kind::Json),
_ if mime.ends_with("+json") => Some(Kind::Json),
_ => None,
}
}
fn extension(path: &str) -> Option<String> {
let path = path.split('?').next().unwrap_or(path);
let segment = path.rsplit('/').next().unwrap_or(path);
let (_, extension) = segment.rsplit_once('.')?;
(!extension.is_empty()).then(|| extension.to_ascii_lowercase())
}
fn extension_kind(path: &str, settings: &Settings) -> Option<Kind> {
let extension = extension(path)?;
if !settings
.extensions
.iter()
.any(|allowed| allowed.eq_ignore_ascii_case(&extension))
{
return None;
}
match extension.as_str() {
"js" | "mjs" | "cjs" => Some(Kind::Javascript),
"css" => Some(Kind::Css),
"html" | "htm" => Some(Kind::Html),
"svg" => Some(Kind::Svg),
"json" => Some(Kind::Json),
_ => None,
}
}
fn kind_enabled(kind: Kind, settings: &Settings) -> bool {
match kind {
Kind::Javascript => settings.javascript,
Kind::Css => settings.css,
Kind::Html => settings.html,
Kind::Svg => settings.svg,
Kind::Json => settings.json,
}
}
fn supported_encoding(headers: &HeaderMap) -> Option<Option<&'static str>> {
let Some(value) = headers.get("content-encoding") else {
return Some(None);
};
let value = value.to_str().ok()?.trim();
if value.eq_ignore_ascii_case("identity") {
return Some(None);
}
for known in ["gzip", "deflate", "br", "zstd"] {
if value.eq_ignore_ascii_case(known) {
return Some(Some(known));
}
}
None
}
fn decode<'a>(body: &'a [u8], encoding: Option<&str>, limit: usize) -> Result<Cow<'a, [u8]>> {
fn read_bounded(mut reader: impl Read, limit: usize, label: &'static str) -> Result<Vec<u8>> {
let mut output = Vec::with_capacity(limit.min(64 * 1024));
let mut chunk = [0u8; 16 * 1024];
loop {
let read = reader
.read(&mut chunk)
.with_context(|| format!("decode {label}"))?;
if read == 0 {
break;
}
ensure!(
output.len().saturating_add(read) <= limit,
"decoded transform body exceeds configured limit"
);
if output.capacity().saturating_sub(output.len()) < read {
output
.try_reserve_exact(read)
.map_err(|_| anyhow::anyhow!("decoded transform allocation failed"))?;
}
output.extend_from_slice(&chunk[..read]);
}
Ok(output)
}
match encoding {
None => {
ensure!(
body.len() <= limit,
"transform body exceeds configured limit"
);
// Identity bodies are already owned by `Bytes` after collection.
// Borrow them instead of creating a second full-size allocation.
Ok(Cow::Borrowed(body))
}
Some("gzip") => {
read_bounded(flate2::read::GzDecoder::new(body), limit, "gzip").map(Cow::Owned)
}
Some("deflate") => {
let zlib = flate2::read::ZlibDecoder::new(body);
match read_bounded(zlib, limit, "deflate") {
Ok(value) => Ok(Cow::Owned(value)),
Err(_) => read_bounded(flate2::read::DeflateDecoder::new(body), limit, "deflate")
.map(Cow::Owned),
}
}
Some("br") => read_bounded(brotli::Decompressor::new(body, 16 * 1024), limit, "brotli")
.map(Cow::Owned),
Some("zstd") => {
let decoder = zstd::stream::read::Decoder::new(body).context("open zstd decoder")?;
read_bounded(decoder, limit, "zstd").map(Cow::Owned)
}
Some(_) => unreachable!(),
}
}
fn drop_vary_accept_encoding(headers: &mut HeaderMap) {
let values: Vec<_> = headers.get_all("vary").iter().cloned().collect();
if values.is_empty() {
return;
}
let mut retained = Vec::<String>::new();
for value in values {
let Ok(value) = value.to_str() else {
return;
};
for token in value
.split(',')
.map(str::trim)
.filter(|token| !token.is_empty())
{
if token == "*" {
// `Vary: *` has special semantics and is rejected by the shared
// cache. Do not rewrite it into a different policy.
return;
}
if token.eq_ignore_ascii_case("accept-encoding") {
continue;
}
if !retained
.iter()
.any(|existing| existing.eq_ignore_ascii_case(token))
{
retained.push(token.to_owned());
}
}
}
headers.remove("vary");
if !retained.is_empty() {
if let Ok(value) = HeaderValue::from_str(&retained.join(", ")) {
headers.insert("vary", value);
}
}
}
fn transformed_etag(body: &[u8]) -> HeaderValue {
let digest = Sha256::digest(body);
let mut value = String::with_capacity(76);
value.push_str("\"rustproxy-");
for byte in digest {
use std::fmt::Write as _;
let _ = write!(&mut value, "{byte:02x}");
}
value.push('"');
HeaderValue::from_str(&value).expect("sha256 validator is always a valid header")
}
fn transform_memory_required_bytes(
content_length: usize,
encoding: Option<&str>,
max_body_bytes: usize,
) -> Option<usize> {
let estimated = if encoding.is_none() {
content_length.checked_mul(2)?
} else {
content_length.checked_add(max_body_bytes.checked_mul(2)?)?
};
let units =
estimated.checked_add(TRANSFORM_MEMORY_UNIT_BYTES - 1)? / TRANSFORM_MEMORY_UNIT_BYTES;
units.max(1).checked_mul(TRANSFORM_MEMORY_UNIT_BYTES)
}
struct TransformMemoryReservation<'a> {
counter: &'a AtomicUsize,
bytes: usize,
}
impl Drop for TransformMemoryReservation<'_> {
fn drop(&mut self) {
self.counter.fetch_sub(self.bytes, Ordering::AcqRel);
}
}
fn try_reserve_memory_from(
counter: &AtomicUsize,
required_bytes: usize,
budget_bytes: usize,
) -> Option<TransformMemoryReservation<'_>> {
if required_bytes > budget_bytes {
return None;
}
let mut current = counter.load(Ordering::Acquire);
loop {
let next = current.checked_add(required_bytes)?;
if next > budget_bytes {
return None;
}
match counter.compare_exchange_weak(current, next, Ordering::AcqRel, Ordering::Acquire) {
Ok(_) => {
return Some(TransformMemoryReservation {
counter,
bytes: required_bytes,
})
}
Err(actual) => current = actual,
}
}
}
fn try_reserve_transform_memory(
required_bytes: usize,
budget_bytes: usize,
) -> Option<TransformMemoryReservation<'static>> {
try_reserve_memory_from(&TRANSFORM_MEMORY_IN_USE_BYTES, required_bytes, budget_bytes)
}
struct TransformInflightReservation<'a> {
counter: &'a AtomicUsize,
}
impl Drop for TransformInflightReservation<'_> {
fn drop(&mut self) {
self.counter.fetch_sub(1, Ordering::AcqRel);
}
}
fn try_reserve_slot_from(
counter: &AtomicUsize,
max_inflight: usize,
) -> Option<TransformInflightReservation<'_>> {
let mut current = counter.load(Ordering::Acquire);
loop {
if current >= max_inflight {
return None;
}
let next = current.checked_add(1)?;
match counter.compare_exchange_weak(current, next, Ordering::AcqRel, Ordering::Acquire) {
Ok(_) => return Some(TransformInflightReservation { counter }),
Err(actual) => current = actual,
}
}
}
fn try_reserve_transform_slot(
max_inflight: usize,
) -> Option<TransformInflightReservation<'static>> {
try_reserve_slot_from(&TRANSFORM_INFLIGHT, max_inflight)
}