itsi-scheduler 0.1.5 → 0.1.19

data/ext/itsi_server/src/server/middleware_stack/middlewares/proxy.rs

@@ -0,0 +1,414 @@
+use std::{
+    collections::HashMap,
+    convert::Infallible,
+    error::Error,
+    net::SocketAddr,
+    sync::{
+        atomic::{AtomicUsize, Ordering},
+        Arc, LazyLock, OnceLock,
+    },
+    time::Duration,
+};
+
+use super::{string_rewrite::StringRewrite, ErrorResponse, FromValue, MiddlewareLayer};
+use crate::{
+    server::{
+        binds::bind::{Bind, BindAddress},
+        http_message_types::{HttpRequest, HttpResponse, RequestExt, ResponseFormat},
+        size_limited_incoming::MaxBodySizeReached,
+    },
+    services::itsi_http_service::HttpRequestContext,
+};
+use async_trait::async_trait;
+use bytes::{Bytes, BytesMut};
+use either::Either;
+use futures::TryStreamExt;
+use http::{HeaderMap, Method, Response, StatusCode};
+use http_body_util::{combinators::BoxBody, BodyExt, Empty, StreamBody};
+use hyper::body::Frame;
+use magnus::error::Result;
+use rand::Rng;
+use reqwest::{
+    dns::{Name, Resolve},
+    Body, Client, Url,
+};
+use serde::Deserialize;
+
+#[derive(Debug, Clone, Deserialize)]
+pub struct Proxy {
+    pub to: StringRewrite,
+    pub backends: Vec<String>,
+    pub backend_priority: BackendPriority,
+    pub headers: HashMap<String, Option<ProxiedHeader>>,
+    pub verify_ssl: bool,
+    pub timeout: u64,
+    pub tls_sni: bool,
+    #[serde(skip_deserializing)]
+    pub client: OnceLock<Client>,
+    #[serde(default = "bad_gateway_error_response")]
+    pub error_response: ErrorResponse,
+}
+
+fn bad_gateway_error_response() -> ErrorResponse {
+    ErrorResponse::bad_gateway()
+}
+
+#[derive(Debug, Clone, Deserialize)]
+pub enum BackendPriority {
+    #[serde(rename(deserialize = "round_robin"))]
+    RoundRobin,
+    #[serde(rename(deserialize = "ordered"))]
+    Ordered,
+    #[serde(rename(deserialize = "random"))]
+    Random,
+}
+
+#[derive(Debug, Clone, Deserialize)]
+pub enum ProxiedHeader {
+    #[serde(rename(deserialize = "value"))]
+    String(String),
+    #[serde(rename(deserialize = "rewrite"))]
+    StringRewrite(StringRewrite),
+}
+
+impl ProxiedHeader {
+    pub fn to_string(&self, req: &HttpRequest, context: &HttpRequestContext) -> String {
+        match self {
+            ProxiedHeader::String(value) => value.clone(),
+            ProxiedHeader::StringRewrite(rewrite) => rewrite.rewrite_request(req, context),
+        }
+    }
+}
+
+#[derive(Debug, Clone)]
+pub struct Resolver {
+    backends: Arc<Vec<SocketAddr>>,
+    counter: Arc<AtomicUsize>,
+    backend_priority: BackendPriority,
+}
+
+pub struct StatefulResolverIter {
+    backends: Arc<Vec<SocketAddr>>,
+    start_index: usize,
+    current: usize,
+}
+
+impl Iterator for StatefulResolverIter {
+    type Item = SocketAddr;
+
+    fn next(&mut self) -> Option<Self::Item> {
+        if self.current < self.backends.len() {
+            let index = (self.start_index + self.current) % self.backends.len();
+            self.current += 1;
+            Some(self.backends[index])
+        } else {
+            None
+        }
+    }
+}
+
+impl Resolve for Resolver {
+    fn resolve(&self, _name: Name) -> reqwest::dns::Resolving {
+        let backends = self.backends.clone();
+        let len = backends.len();
+
+        let start_index = match self.backend_priority {
+            BackendPriority::Ordered => 0,
+            BackendPriority::Random => rand::rng().random_range(0..len),
+            BackendPriority::RoundRobin => self.counter.fetch_add(1, Ordering::Relaxed) % len,
+        };
+
+        let fut = async move {
+            let iter = StatefulResolverIter {
+                backends,
+                start_index,
+                current: 0,
+            };
+            Ok(Box::new(iter) as Box<dyn Iterator<Item = SocketAddr> + Send>)
+        };
+
+        Box::pin(fut)
+    }
+}
+
+static BAD_GATEWAY_RESPONSE: LazyLock<ErrorResponse> = LazyLock::new(ErrorResponse::bad_gateway);
+static GATEWAY_TIMEOUT_RESPONSE: LazyLock<ErrorResponse> =
+    LazyLock::new(ErrorResponse::gateway_timeout);
+static SERVICE_UNAVAILABLE_RESPONSE: LazyLock<ErrorResponse> =
+    LazyLock::new(ErrorResponse::service_unavailable);
+static INTERNAL_SERVER_ERROR_RESPONSE: LazyLock<ErrorResponse> =
+    LazyLock::new(ErrorResponse::internal_server_error);
+
+fn is_idempotent(method: &Method) -> bool {
+    matches!(
+        *method,
+        Method::GET | Method::HEAD | Method::PUT | Method::DELETE | Method::OPTIONS
+    )
+}
+
+/// A helper that stores the immutable parts of the incoming request.
+struct RequestInfo {
+    method: Method,
+    headers: HeaderMap,
+}
+
+impl Proxy {
+    /// Build a header map of overriding headers based on the configured values.
+    /// This uses the full HttpRequest and context to compute each header value.
+    fn build_overriding_headers(
+        &self,
+        req: &HttpRequest,
+        context: &mut HttpRequestContext,
+    ) -> http::HeaderMap {
+        let mut headers = http::HeaderMap::new();
+        for (name, header_opt) in self.headers.iter() {
+            if let Some(header_value) = header_opt {
+                // Compute the header value using the full HttpRequest.
+                let value_str = header_value.to_string(req, context);
+                if let Ok(header_val) = http::HeaderValue::from_str(&value_str) {
+                    if let Ok(header_name) = name.parse::<http::header::HeaderName>() {
+                        headers.insert(header_name, header_val);
+                    }
+                }
+            }
+        }
+        headers
+    }
+
+    /// Build a reqwest::RequestBuilder by merging the original request headers
+    /// (unless overridden) with the precomputed overriding headers.
+    fn build_reqwest_request_info(
+        &self,
+        req_info: &RequestInfo,
+        url: &str,
+        host_str: &str,
+        body: Body,
+        overriding_headers: &http::HeaderMap,
+    ) -> reqwest::RequestBuilder {
+        let mut builder = self
+            .client
+            .get()
+            .unwrap()
+            .request(req_info.method.clone(), url);
+
+        // Forward headers from the original request unless they are overridden.
+        for (name, value) in req_info.headers.iter() {
+            if overriding_headers.contains_key(name) {
+                continue;
+            }
+            builder = builder.header(name, value);
+        }
+
+        // Add a Host header if not overridden.
+        if !overriding_headers.contains_key("host") && !host_str.is_empty() {
+            builder = builder.header("Host", host_str);
+        }
+
+        for (name, value) in overriding_headers.iter() {
+            builder = builder.header(name, value);
+        }
+
+        builder.body(body)
+    }
+
+    /// Sends an idempotent request using a replayable (buffered) body.
+    async fn send_request_idempotent(
+        &self,
+        req_info: &RequestInfo,
+        url: &str,
+        host_str: &str,
+        max_attempts: usize,
+        replayable_bytes: Bytes,
+        overriding_headers: &http::HeaderMap,
+    ) -> std::result::Result<reqwest::Response, reqwest::Error> {
+        let mut last_err = None;
+        for attempt in 0..max_attempts {
+            let body = Body::from(replayable_bytes.clone());
+            let builder =
+                self.build_reqwest_request_info(req_info, url, host_str, body, overriding_headers);
+            match builder.send().await {
+                Ok(response) => return Ok(response),
+                Err(e) => {
+                    // Retry for connectivity-related errors.
+                    if e.is_connect() {
+                        last_err = Some(e);
+                        if attempt + 1 < max_attempts {
+                            continue;
+                        } else {
+                            break;
+                        }
+                    } else {
+                        return Err(e);
+                    }
+                }
+            }
+        }
+        Err(last_err.expect("At least one attempt should have set last_err"))
+    }
+
+    /// Sends a non-idempotent request once using its streaming body.
+    async fn send_request_non_idempotent(
+        &self,
+        req: HttpRequest,
+        req_info: &RequestInfo,
+        url: &str,
+        host_str: &str,
+        overriding_headers: &http::HeaderMap,
+    ) -> std::result::Result<reqwest::Response, reqwest::Error> {
+        let body = Body::wrap_stream(req.into_data_stream());
+        let builder =
+            self.build_reqwest_request_info(req_info, url, host_str, body, overriding_headers);
+        builder.send().await
+    }
+}
+
+#[async_trait]
+impl MiddlewareLayer for Proxy {
+    async fn initialize(&self) -> Result<()> {
+        let backends = self
+            .backends
+            .iter()
+            .filter_map(|be| {
+                let bind: Bind = be.parse().ok()?;
+                match (bind.address, bind.port) {
+                    (BindAddress::Ip(ip_addr), port) => {
+                        Some(SocketAddr::new(ip_addr, port.unwrap()))
+                    }
+                    (BindAddress::UnixSocket(_), _) => None,
+                }
+            })
+            .collect::<Vec<_>>();
+
+        self.client
+            .set(
+                Client::builder()
+                    .timeout(Duration::from_secs(self.timeout))
+                    .danger_accept_invalid_certs(!self.verify_ssl)
+                    .danger_accept_invalid_hostnames(!self.verify_ssl)
+                    .dns_resolver(Arc::new(Resolver {
+                        backends: Arc::new(backends),
+                        counter: Arc::new(AtomicUsize::new(0)),
+                        backend_priority: self.backend_priority.clone(),
+                    }))
+                    .tls_sni(self.tls_sni)
+                    .build()
+                    .map_err(|e| {
+                        magnus::Error::new(
+                            magnus::exception::runtime_error(),
+                            format!("Failed to build Reqwest client: {}", e),
+                        )
+                    })?,
+            )
+            .map_err(|_e| {
+                magnus::Error::new(
+                    magnus::exception::standard_error(),
+                    "Failed to save resolver backends",
+                )
+            })?;
+        Ok(())
+    }
+
+    async fn before(
+        &self,
+        req: HttpRequest,
+        context: &mut HttpRequestContext,
+    ) -> Result<Either<HttpRequest, HttpResponse>> {
+        let url = self.to.rewrite_request(&req, context);
+        let accept: ResponseFormat = req.accept().into();
+        let error_response = self.error_response.to_http_response(accept.clone()).await;
+
+        let destination = match Url::parse(&url) {
+            Ok(dest) => dest,
+            Err(_) => return Ok(Either::Right(error_response)),
+        };
+
+        // Clone the headers before consuming the request.
+        let req_headers = req.headers().clone();
+        let host_str = destination.host_str().unwrap_or_else(|| {
+            req_headers
+                .get("Host")
+                .and_then(|h| h.to_str().ok())
+                .unwrap_or("")
+        });
+
+        let req_info = RequestInfo {
+            method: req.method().clone(),
+            headers: req_headers.clone(),
+        };
+
+        // Precompute the overriding headers from the full request.
+        let overriding_headers = self.build_overriding_headers(&req, context);
+
+        // Determine max_attempts based on the number of backends.
+        let max_attempts = self.backends.len();
+
+        let reqwest_response_result = if is_idempotent(&req_info.method) {
+            let (_parts, body) = req.into_parts();
+            let replayable_bytes = match body.into_data_stream().try_collect::<Vec<Bytes>>().await {
+                Ok(chunks) => {
+                    let mut buf = BytesMut::new();
+                    for chunk in chunks {
+                        buf.extend_from_slice(&chunk);
+                    }
+                    buf.freeze()
+                }
+                Err(e) => {
+                    tracing::error!("Error buffering request body: {}", e);
+                    return Ok(Either::Right(error_response));
+                }
+            };
+            self.send_request_idempotent(
+                &req_info,
+                &url,
+                host_str,
+                max_attempts,
+                replayable_bytes,
+                &overriding_headers,
+            )
+            .await
+        } else {
+            self.send_request_non_idempotent(req, &req_info, &url, host_str, &overriding_headers)
+                .await
+        };
+
+        let response = match reqwest_response_result {
+            Ok(response) => {
+                let status = response.status();
+                let mut builder = Response::builder().status(status);
+                for (hn, hv) in response.headers() {
+                    builder = builder.header(hn, hv);
+                }
+                let response = builder.body(BoxBody::new(StreamBody::new(
+                    response
+                        .bytes_stream()
+                        .map_ok(Frame::data)
+                        .map_err(|_| -> Infallible { unreachable!("We handle IO errors above") }),
+                )));
+                response.unwrap_or(error_response)
+            }
+            Err(e) => {
+                if let Some(inner) = e.source() {
+                    if inner.downcast_ref::<MaxBodySizeReached>().is_some() {
+                        let mut max_body_response = Response::new(BoxBody::new(Empty::new()));
+                        *max_body_response.status_mut() = StatusCode::PAYLOAD_TOO_LARGE;
+                        return Ok(Either::Right(max_body_response));
+                    }
+                }
+                if e.is_timeout() {
+                    GATEWAY_TIMEOUT_RESPONSE.to_http_response(accept).await
+                } else if e.is_connect() {
+                    BAD_GATEWAY_RESPONSE.to_http_response(accept).await
+                } else if e.is_status() {
+                    SERVICE_UNAVAILABLE_RESPONSE.to_http_response(accept).await
+                } else {
+                    INTERNAL_SERVER_ERROR_RESPONSE
+                        .to_http_response(accept)
+                        .await
+                }
+            }
+        };
+
+        Ok(Either::Right(response))
+    }
+}
+impl FromValue for Proxy {}

data/ext/itsi_server/src/server/middleware_stack/middlewares/rate_limit.rs

@@ -0,0 +1,131 @@
+use super::{token_source::TokenSource, ErrorResponse, FromValue, MiddlewareLayer};
+use crate::server::http_message_types::{HttpRequest, HttpResponse, RequestExt};
+use crate::services::itsi_http_service::HttpRequestContext;
+use crate::services::rate_limiter::{
+    create_rate_limit_key, get_rate_limiter, RateLimitError, RateLimiter, RateLimiterConfig,
+};
+use async_trait::async_trait;
+use either::Either;
+use magnus::error::Result;
+use serde::Deserialize;
+use std::sync::{Arc, OnceLock};
+use std::time::Duration;
+
+#[derive(Debug, Clone, Deserialize)]
+pub struct RateLimit {
+    pub requests: u64,
+    pub seconds: u64,
+    pub key: RateLimitKey,
+    #[serde(skip_deserializing)]
+    pub rate_limiter: OnceLock<Arc<dyn RateLimiter>>,
+    pub store_config: RateLimiterConfig,
+    #[serde(default = "too_many_requests_error_response")]
+    pub error_response: ErrorResponse,
+}
+
+fn too_many_requests_error_response() -> ErrorResponse {
+    ErrorResponse::too_many_requests()
+}
+
+#[derive(Debug, Clone, Deserialize)]
+pub enum RateLimitKey {
+    #[serde(rename(deserialize = "address"))]
+    SocketAddress,
+    #[serde(rename(deserialize = "parameter"))]
+    Parameter(TokenSource),
+}
+
+#[async_trait]
+impl MiddlewareLayer for RateLimit {
+    async fn initialize(&self) -> Result<()> {
+        // Instantiate our rate limiter based on the rate limit config here.
+        // This will automatically fall back to in-memory if Redis fails
+        if let Ok(limiter) = get_rate_limiter(&self.store_config).await {
+            let _ = self.rate_limiter.set(limiter);
+        }
+        Ok(())
+    }
+
+    async fn before(
+        &self,
+        req: HttpRequest,
+        context: &mut HttpRequestContext,
+    ) -> Result<Either<HttpRequest, HttpResponse>> {
+        // Get the key to rate limit on
+        let key_value = match &self.key {
+            RateLimitKey::SocketAddress => {
+                // Use the socket address from the context
+                &context.addr
+            }
+            RateLimitKey::Parameter(token_source) => {
+                match token_source {
+                    TokenSource::Header { name, prefix } => {
+                        if let Some(header) = req.header(name) {
+                            if let Some(prefix) = prefix {
+                                header.strip_prefix(prefix).unwrap_or("").trim_ascii()
+                            } else {
+                                header.trim_ascii()
+                            }
+                        } else {
+                            // If no token is found, skip rate limiting
+                            tracing::warn!("No token found in header for rate limiting");
+                            return Ok(Either::Left(req));
+                        }
+                    }
+                    TokenSource::Query(query_name) => {
+                        if let Some(value) = req.query_param(query_name) {
+                            value
+                        } else {
+                            // If no token is found, skip rate limiting
+                            tracing::warn!("No token found in query for rate limiting");
+                            return Ok(Either::Left(req));
+                        }
+                    }
+                }
+            }
+        };
+
+        // Create a rate limit key
+        let rate_limit_key = create_rate_limit_key(key_value, req.uri().path());
+
+        // Get the rate limiter
+        if let Some(limiter) = self.rate_limiter.get() {
+            // Check if rate limit is exceeded
+            let timeout = Duration::from_secs(self.seconds);
+            let limit = self.requests;
+
+            match limiter.check_limit(&rate_limit_key, limit, timeout).await {
+                Ok(_) => Ok(Either::Left(req)),
+                Err(RateLimitError::RateLimitExceeded { limit, ttl, .. }) => {
+                    let mut response = self
+                        .error_response
+                        .to_http_response(req.accept().into())
+                        .await;
+                    response
+                        .headers_mut()
+                        .insert("X-RateLimit-Limit", limit.to_string().parse().unwrap());
+                    response
+                        .headers_mut()
+                        .insert("X-RateLimit-Remaining", "0".parse().unwrap());
+                    response
+                        .headers_mut()
+                        .insert("X-RateLimit-Reset", ttl.to_string().parse().unwrap());
+                    response
+                        .headers_mut()
+                        .insert("Retry-After", ttl.to_string().parse().unwrap());
+                    Ok(Either::Right(response))
+                }
+                Err(e) => {
+                    // Other error, log and allow request (fail open)
+                    tracing::error!("Rate limiter error: {:?}", e);
+                    Ok(Either::Left(req))
+                }
+            }
+        } else {
+            // If rate limiter is not initialized, allow request
+            tracing::warn!("Rate limiter not initialized");
+            Ok(Either::Left(req))
+        }
+    }
+}
+impl FromValue for RateLimit {}

data/ext/itsi_server/src/server/middleware_stack/middlewares/redirect.rs

@@ -0,0 +1,76 @@
+use crate::{
+    server::http_message_types::{HttpRequest, HttpResponse},
+    services::itsi_http_service::HttpRequestContext,
+};
+
+use super::{string_rewrite::StringRewrite, FromValue, MiddlewareLayer};
+
+use async_trait::async_trait;
+use either::Either;
+use http::{Response, StatusCode};
+use http_body_util::{combinators::BoxBody, Empty};
+use magnus::error::Result;
+use serde::Deserialize;
+
+/// A simple API key filter.
+/// The API key can be given inside the header or a query string
+/// Keys are validated against a list of allowed key values (Changing these requires a restart)
+///
+#[derive(Debug, Clone, Deserialize)]
+pub struct Redirect {
+    pub to: StringRewrite,
+    #[serde(default)]
+    #[serde(rename(deserialize = "type"))]
+    pub redirect_type: RedirectType,
+}
+
+#[derive(Debug, Clone, Deserialize, Default)]
+pub enum RedirectType {
+    #[serde(rename(deserialize = "permanent"))]
+    #[default]
+    Permanent,
+    #[serde(rename(deserialize = "temporary"))]
+    Temporary,
+    #[serde(rename(deserialize = "found"))]
+    Found,
+    #[serde(rename(deserialize = "moved_permanently"))]
+    MovedPermanently,
+}
+
+#[async_trait]
+impl MiddlewareLayer for Redirect {
+    async fn before(
+        &self,
+        req: HttpRequest,
+        context: &mut HttpRequestContext,
+    ) -> Result<Either<HttpRequest, HttpResponse>> {
+        Ok(Either::Right(self.redirect_response(&req, context)?))
+    }
+}
+
+impl Redirect {
+    pub fn redirect_response(
+        &self,
+        req: &HttpRequest,
+        context: &mut HttpRequestContext,
+    ) -> Result<HttpResponse> {
+        let mut response = Response::new(BoxBody::new(Empty::new()));
+        *response.status_mut() = match self.redirect_type {
+            RedirectType::Permanent => StatusCode::PERMANENT_REDIRECT,
+            RedirectType::Temporary => StatusCode::TEMPORARY_REDIRECT,
+            RedirectType::MovedPermanently => StatusCode::MOVED_PERMANENTLY,
+            RedirectType::Found => StatusCode::FOUND,
+        };
+        response.headers_mut().append(
+            "Location",
+            self.to.rewrite_request(req, context).parse().map_err(|e| {
+                magnus::Error::new(
+                    magnus::exception::standard_error(),
+                    format!("Invalid Rewrite String: {:?}: {}", self.to, e),
+                )
+            })?,
+        );
+        Ok(response)
+    }
+}
+impl FromValue for Redirect {}

data/ext/itsi_server/src/server/middleware_stack/middlewares/request_headers.rs

@@ -0,0 +1,44 @@
+use std::collections::HashMap;
+
+use crate::{
+    server::http_message_types::{HttpRequest, HttpResponse},
+    services::itsi_http_service::HttpRequestContext,
+};
+
+use super::{FromValue, MiddlewareLayer};
+use async_trait::async_trait;
+use either::Either;
+use http::HeaderName;
+use magnus::error::Result;
+use serde::Deserialize;
+
+#[derive(Debug, Clone, Deserialize)]
+pub struct RequestHeaders {
+    pub additions: HashMap<String, Vec<String>>,
+    pub removals: Vec<String>,
+}
+
+#[async_trait]
+impl MiddlewareLayer for RequestHeaders {
+    async fn before(
+        &self,
+        mut req: HttpRequest,
+        _: &mut HttpRequestContext,
+    ) -> Result<Either<HttpRequest, HttpResponse>> {
+        let headers = req.headers_mut();
+        for removal in &self.removals {
+            headers.remove(removal);
+        }
+        for (header_name, header_values) in &self.additions {
+            for header_value in header_values {
+                if let Ok(parsed_header_name) = header_name.parse::<HeaderName>() {
+                    if let Ok(parsed_header_value) = header_value.parse() {
+                        headers.append(parsed_header_name, parsed_header_value);
+                    }
+                }
+            }
+        }
+        Ok(Either::Left(req))
+    }
+}
+impl FromValue for RequestHeaders {}

data/ext/itsi_server/src/server/middleware_stack/middlewares/response_headers.rs

@@ -0,0 +1,36 @@
+use std::collections::HashMap;
+
+use super::{FromValue, MiddlewareLayer};
+use crate::{
+    server::http_message_types::HttpResponse, services::itsi_http_service::HttpRequestContext,
+};
+use async_trait::async_trait;
+use http::HeaderName;
+use serde::Deserialize;
+
+#[derive(Debug, Clone, Deserialize)]
+pub struct ResponseHeaders {
+    pub additions: HashMap<String, Vec<String>>,
+    pub removals: Vec<String>,
+}
+
+#[async_trait]
+impl MiddlewareLayer for ResponseHeaders {
+    async fn after(&self, mut resp: HttpResponse, _: &mut HttpRequestContext) -> HttpResponse {
+        let headers = resp.headers_mut();
+        for removal in &self.removals {
+            headers.remove(removal);
+        }
+        for (header_name, header_values) in &self.additions {
+            for header_value in header_values {
+                if let Ok(parsed_header_name) = header_name.parse::<HeaderName>() {
+                    if let Ok(parsed_header_value) = header_value.parse() {
+                        headers.append(parsed_header_name, parsed_header_value);
+                    }
+                }
+            }
+        }
+        resp
+    }
+}
+impl FromValue for ResponseHeaders {}