use super::{JsonRpcParams, LooseId, SingleRequest}; use crate::{ app::App, block_number::CacheMode, errors::Web3ProxyResult, frontend::{ authorization::{key_is_authorized, Authorization, RequestOrMethod, ResponseOrBytes}, rpc_proxy_ws::ProxyMode, }, globals::APP, response_cache::JsonRpcQueryCacheKey, rpcs::{blockchain::Web3ProxyBlock, one::Web3Rpc}, secrets::RpcSecretKey, stats::{AppStat, BackendRequests}, }; use anyhow::Context; use axum::headers::{Origin, Referer, UserAgent}; use chrono::Utc; use derivative::Derivative; use ethers::types::U64; use rust_decimal::Decimal; use serde::{ser::SerializeStruct, Serialize}; use serde_json::{json, value::RawValue}; use std::{borrow::Cow, sync::Arc}; use std::{ fmt::{self, Display}, net::IpAddr, }; use std::{ mem, sync::atomic::{self, AtomicBool, AtomicI64, AtomicU64}, time::Duration, }; use tokio::{ sync::{mpsc, OwnedSemaphorePermit}, time::Instant, }; use tracing::{error, trace}; #[cfg(feature = "rdkafka")] use { crate::{jsonrpc, kafka::KafkaDebugLogger}, tracing::warn, ulid::Ulid, }; #[derive(Derivative)] #[derivative(Default)] pub struct RequestBuilder { app: Option>, archive_request: bool, head_block: Option, authorization: Option>, request_or_method: RequestOrMethod, } impl RequestBuilder { pub fn new(app: Arc) -> Self { let head_block = app.head_block_receiver().borrow().clone(); Self { app: Some(app), head_block, ..Default::default() } } pub fn authorize_internal(self) -> Web3ProxyResult { // TODO: allow passing proxy_mode to internal? let authorization = Authorization::internal()?; Ok(Self { authorization: Some(Arc::new(authorization)), ..self }) } pub async fn authorize_premium( self, ip: &IpAddr, rpc_key: &RpcSecretKey, origin: Option<&Origin>, user_agent: Option<&UserAgent>, proxy_mode: ProxyMode, referer: Option<&Referer>, ) -> Web3ProxyResult { let app = self .app .as_ref() .context("app is required for public requests")?; // TODO: we can check authorization now, but the semaphore needs to wait! let authorization = key_is_authorized(app, rpc_key, ip, origin, proxy_mode, referer, user_agent).await?; Ok(Self { authorization: Some(Arc::new(authorization)), ..self }) } /// TODO: this takes a lot more things pub fn authorize_public( self, ip: &IpAddr, origin: Option<&Origin>, user_agent: Option<&UserAgent>, proxy_mode: ProxyMode, referer: Option<&Referer>, ) -> Web3ProxyResult { let app = self .app .as_ref() .context("app is required for public requests")?; let authorization = Authorization::external( &app.config.allowed_origin_requests_per_period, ip, origin, proxy_mode, referer, user_agent, )?; let head_block = app.watch_consensus_head_receiver.borrow().clone(); Ok(Self { authorization: Some(Arc::new(authorization)), head_block, ..self }) } pub fn set_archive_request(self, archive_request: bool) -> Self { Self { archive_request, ..self } } /// replace 'latest' in the json and figure out the minimum and maximum blocks. /// also tarpit invalid methods. pub async fn set_request(self, request: SingleRequest) -> Web3ProxyResult { Ok(Self { request_or_method: RequestOrMethod::Request(request), ..self }) } pub fn set_method(self, method: Cow<'static, str>, size: usize) -> Self { Self { request_or_method: RequestOrMethod::Method(method, size), ..self } } pub async fn build( &self, max_wait: Option, ) -> Web3ProxyResult> { // TODO: make this work without app being set? let app = self.app.as_ref().context("app is required")?; let authorization = self .authorization .clone() .context("authorization is required")?; let permit = authorization.permit(app).await?; let x = ValidatedRequest::new_with_app( app, authorization, max_wait, permit, self.request_or_method.clone(), self.head_block.clone(), ) .await; if let Ok(x) = &x { if self.archive_request { x.archive_request.store(true, atomic::Ordering::Relaxed); } } // todo!(anything else to set?) x } } /// TODO: /// TODO: instead of a bunch of atomics, this should probably use a RwLock. need to think more about how parallel requests are going to work though #[derive(Debug, Derivative)] #[derivative(Default)] pub struct ValidatedRequest { /// TODO: set archive_request during the new instead of after /// TODO: this is more complex than "requires a block older than X height". different types of data can be pruned differently pub archive_request: AtomicBool, pub authorization: Arc, pub cache_mode: CacheMode, /// TODO: this should probably be in a global config. although maybe if we run multiple chains in one process this will be useful pub chain_id: u64, pub head_block: Option, /// TODO: this should be in a global config. not copied to every single request pub usd_per_cu: Decimal, pub inner: RequestOrMethod, /// Instant that the request was received (or at least close to it) /// We use Instant and not timestamps to avoid problems with leap seconds and similar issues #[derivative(Default(value = "Instant::now()"))] pub start_instant: Instant, /// if this is empty, there was a cache_hit /// otherwise, it is populated with any rpc servers that were used by this request pub backend_requests: BackendRequests, /// The number of times the request got stuck waiting because no servers were synced pub no_servers: AtomicU64, /// If handling the request hit an application error /// This does not count things like a transcation reverting or a malformed request pub error_response: AtomicBool, /// Size in bytes of the JSON response. Does not include headers or things like that. pub response_bytes: AtomicU64, /// How many milliseconds it took to respond to the request pub response_millis: AtomicU64, /// What time the (first) response was proxied. /// TODO: think about how to store response times for ProxyMode::Versus pub response_timestamp: AtomicI64, /// True if the response required querying a backup RPC /// RPC aggregators that query multiple providers to compare response may use this header to ignore our response. pub response_from_backup_rpc: AtomicBool, /// If the request is invalid or received a jsonrpc error response (excluding reverts) pub user_error_response: AtomicBool, #[cfg(feature = "rdkafka")] /// ProxyMode::Debug logs requests and responses with Kafka /// TODO: maybe this shouldn't be determined by ProxyMode. A request param should probably enable this pub kafka_debug_logger: Option>, #[cfg(not(feature = "rdkafka"))] pub kafka_debug_logger: Option<()>, /// Cancel-safe channel for sending stats to the buffer pub stat_sender: Option>, /// How long to spend waiting for an rpc that can serve this request pub connect_timeout: Duration, /// How long to spend waiting for an rpc to respond to this request /// TODO: this should start once the connection is established pub expire_timeout: Duration, /// limit the number of concurrent requests from a given user. pub permit: Option, } impl Display for ValidatedRequest { fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> fmt::Result { write!( f, "{}({})", self.inner.method(), serde_json::to_string(self.inner.params()).expect("this should always serialize") ) } } impl Serialize for ValidatedRequest { fn serialize(&self, serializer: S) -> Result where S: serde::Serializer, { let mut state = serializer.serialize_struct("request", 7)?; state.serialize_field( "archive_request", &self.archive_request.load(atomic::Ordering::Relaxed), )?; state.serialize_field("chain_id", &self.chain_id)?; state.serialize_field("head_block", &self.head_block)?; state.serialize_field("request", &self.inner)?; state.serialize_field("elapsed", &self.start_instant.elapsed().as_secs_f32())?; { let backend_names = self.backend_requests.lock(); let backend_names = backend_names .iter() .map(|x| x.name.as_str()) .collect::>(); state.serialize_field("backend_requests", &backend_names)?; } state.serialize_field( "response_bytes", &self.response_bytes.load(atomic::Ordering::Relaxed), )?; state.end() } } /// TODO: move all of this onto RequestBuilder impl ValidatedRequest { #[allow(clippy::too_many_arguments)] async fn new_with_options( app: Option<&App>, authorization: Arc, chain_id: u64, mut head_block: Option, #[cfg(feature = "rdkafka")] kafka_debug_logger: Option>, max_wait: Option, permit: Option, mut request: RequestOrMethod, usd_per_cu: Decimal, ) -> Web3ProxyResult> { let start_instant = Instant::now(); let stat_sender = app.and_then(|x| x.stat_sender.clone()); // let request: RequestOrMethod = request.into(); // we VERY INTENTIONALLY log to kafka BEFORE calculating the cache key // this is because calculating the cache_key may modify the params! // for example, if the request specifies "latest" as the block number, we replace it with the actual latest block number #[cfg(feature = "rdkafka")] if let Some(ref kafka_debug_logger) = kafka_debug_logger { // TODO: channels might be more ergonomic than spawned futures // spawned things run in parallel easier but generally need more Arcs kafka_debug_logger.log_debug_request(&request); } #[cfg(not(feature = "rdkafka"))] let kafka_debug_logger = None; if head_block.is_none() { if let Some(app) = app { head_block = app.head_block_receiver().borrow().clone(); } } // now that kafka has logged the user's original params, we can calculate the cache key // TODO: modify CacheMode::new to wait for a future block if one is requested! be sure to update head_block too! let cache_mode = match &mut request { RequestOrMethod::Request(x) => CacheMode::new(x, head_block.as_ref(), app).await, _ => CacheMode::Never, }; let connect_timeout = Duration::from_secs(3); let expire_timeout = max_wait.unwrap_or_else(|| Duration::from_secs(295)); let x = Self { archive_request: false.into(), authorization, backend_requests: Default::default(), cache_mode, chain_id, error_response: false.into(), connect_timeout, expire_timeout, head_block: head_block.clone(), kafka_debug_logger, no_servers: 0.into(), inner: request, permit, response_bytes: 0.into(), response_from_backup_rpc: false.into(), response_millis: 0.into(), response_timestamp: 0.into(), start_instant, stat_sender, usd_per_cu, user_error_response: false.into(), }; Ok(Arc::new(x)) } /// todo!(this shouldn't be public. use the RequestBuilder) pub async fn new_with_app( app: &App, authorization: Arc, max_wait: Option, permit: Option, request: RequestOrMethod, head_block: Option, ) -> Web3ProxyResult> { #[cfg(feature = "rdkafka")] let kafka_debug_logger = if matches!(authorization.checks.proxy_mode, ProxyMode::Debug) { // TODO: get this out of tracing instead (where we have a String from Amazon's LB) let request_ulid = Ulid::new(); KafkaDebugLogger::try_new( app, authorization.clone(), head_block.as_ref().map(|x| x.number()), "web3_proxy:rpc", request_ulid, ) } else { None }; let chain_id = app.config.chain_id; let usd_per_cu = app.config.usd_per_cu.unwrap_or_default(); Self::new_with_options( Some(app), authorization, chain_id, head_block, #[cfg(feature = "rdkafka")] kafka_debug_logger, max_wait, permit, request, usd_per_cu, ) .await } pub async fn new_internal( method: Cow<'static, str>, params: &P, head_block: Option, max_wait: Option, ) -> Web3ProxyResult> { let authorization = Arc::new(Authorization::internal().unwrap()); // todo!(we need a real id! increment a counter on the app or websocket-only providers are going to have a problem) let id = LooseId::Number(1); // TODO: this seems inefficient let request = SingleRequest::new(id, method, json!(params)).unwrap(); if let Some(app) = APP.get() { Self::new_with_app( app, authorization, max_wait, None, request.into(), head_block, ) .await } else { Self::new_with_options( None, authorization, 0, head_block, #[cfg(feature = "rdkafka")] None, max_wait, None, request.into(), Default::default(), ) .await } } #[inline] pub fn backend_rpcs_used(&self) -> Vec> { self.backend_requests.lock().clone() } pub fn cache_key(&self) -> Option { match &self.cache_mode { CacheMode::Never => None, x => { let x = JsonRpcQueryCacheKey::new(x, &self.inner).hash(); Some(x) } } } #[inline] pub fn cache_jsonrpc_errors(&self) -> bool { self.cache_mode.cache_jsonrpc_errors() } #[inline] pub fn id(&self) -> Box { self.inner.id() } #[inline] pub fn max_block_needed(&self) -> Option { self.cache_mode.to_block().map(|x| *x.num()) } #[inline] pub fn min_block_needed(&self) -> Option { if self.archive_request.load(atomic::Ordering::Relaxed) { Some(U64::zero()) } else { self.cache_mode.from_block().map(|x| *x.num()) } } #[inline] pub fn connect_timeout_at(&self) -> Instant { // TODO: get from config self.start_instant + Duration::from_secs(3) } #[inline] pub fn connect_timeout(&self) -> bool { self.connect_timeout_at() <= Instant::now() } #[inline] pub fn expire_at(&self) -> Instant { // TODO: get from config // erigon's timeout is 5 minutes so we want it shorter than that self.start_instant + Duration::from_secs(295) } #[inline] pub fn expired(&self) -> bool { self.expire_at() <= Instant::now() } pub fn try_send_stat(mut self) -> Web3ProxyResult<()> { if let Some(stat_sender) = self.stat_sender.take() { trace!(?self, "sending stat"); let stat: AppStat = self.into(); if let Err(err) = stat_sender.send(stat) { error!(?err, "failed sending stat"); // TODO: return it? that seems like it might cause an infinite loop // TODO: but dropping stats is bad... hmm... i guess better to undercharge customers than overcharge }; trace!("stat sent successfully"); } Ok(()) } pub fn add_response<'a, R: Into>>(&'a self, response: R) { // TODO: fetch? set? should it be None in a Mutex? or a OnceCell? let response = response.into(); let num_bytes = response.num_bytes(); self.response_bytes .fetch_add(num_bytes, atomic::Ordering::Relaxed); self.response_millis.fetch_add( self.start_instant.elapsed().as_millis() as u64, atomic::Ordering::Relaxed, ); // TODO: record first or last timestamp? really, we need multiple self.response_timestamp .store(Utc::now().timestamp(), atomic::Ordering::Relaxed); // TODO: set user_error_response and error_response here instead of outside this function #[cfg(feature = "rdkafka")] if let Some(kafka_debug_logger) = self.kafka_debug_logger.as_ref() { if let ResponseOrBytes::Response(response) = response { match response { jsonrpc::SingleResponse::Parsed(response) => { kafka_debug_logger.log_debug_response(response); } jsonrpc::SingleResponse::Stream(_) => { warn!("need to handle streaming response debug logging"); } } } } } pub fn try_send_arc_stat(self: Arc) -> Web3ProxyResult<()> { match Arc::into_inner(self) { Some(x) => x.try_send_stat(), None => { trace!("could not send stat while other arcs are active"); Ok(()) } } } #[inline] pub fn proxy_mode(&self) -> ProxyMode { self.authorization.checks.proxy_mode } // TODO: helper function to duplicate? needs to clear request_bytes, and all the atomics tho... } impl Drop for ValidatedRequest { fn drop(&mut self) { if self.stat_sender.is_some() { // turn `&mut self` into `self` let x = mem::take(self); // trace!(?x, "request metadata dropped without stat send"); let _ = x.try_send_stat(); } } } /* pub fn foo(&self) { let payload = payload .map_err(|e| Web3ProxyError::from(e).into_response_with_id(None))? .0; let first_id = payload.first_id(); let (authorization, _semaphore) = ip_is_authorized(&app, ip, origin, proxy_mode) .await .map_err(|e| e.into_response_with_id(first_id.clone()))?; let authorization = Arc::new(authorization); payload .tarpit_invalid(&app, &authorization, Duration::from_secs(5)) .await?; // TODO: calculate payload bytes here (before turning into serde_json::Value). that will save serializing later // TODO: is first_id the right thing to attach to this error? let (status_code, response, rpcs) = app .proxy_web3_rpc(authorization, payload) .await .map_err(|e| e.into_response_with_id(first_id))?; let mut response = (status_code, response).into_response(); // TODO: DRY this up. it is the same code for public and private queries let response_headers = response.headers_mut(); // TODO: this might be slow. think about this more // TODO: special string if no rpcs were used (cache hit)? let mut backup_used = false; let rpcs: String = rpcs .into_iter() .map(|x| { if x.backup { backup_used = true; } x.name.clone() }) .join(","); response_headers.insert( "X-W3P-BACKEND-RPCS", rpcs.parse().expect("W3P-BACKEND-RPCS should always parse"), ); response_headers.insert( "X-W3P-BACKUP-RPC", backup_used .to_string() .parse() .expect("W3P-BACKEND-RPCS should always parse"), ); Ok(response) } */