itsi-server 0.1.1 → 0.1.18

data/ext/itsi_server/src/ruby_types/itsi_body_proxy/big_bytes.rs

@@ -0,0 +1,109 @@
+use bytes::Bytes;
+use magnus::{IntoValue, Ruby, Value};
+use std::io::{Result as IoResult, Write};
+use std::path::PathBuf;
+use tempfile::NamedTempFile;
+
+const THRESHOLD: usize = 1024 * 1024; // 1 MB
+
+/// A container that will hold data in memory if it’s small, or in a temporary file on disk if it exceeds THRESHOLD.
+/// Used for providing Rack input data.
+pub enum BigBytes {
+    InMemory(Vec<u8>),
+    OnDisk(NamedTempFile),
+}
+
+/// The result type for reading the contents of a `BigBytes` value.
+pub enum BigBytesReadResult {
+    /// When the data is stored in memory, returns the cached bytes.
+    InMemory(Vec<u8>),
+    /// When the data is stored on disk, returns the path to the temporary file.
+    OnDisk(PathBuf),
+}
+
+impl BigBytes {
+    /// Creates a new, empty BigBytes instance (initially in memory).
+    pub fn new() -> Self {
+        BigBytes::InMemory(Vec::new())
+    }
+
+    /// Returns either the raw bytes, or the file path of the BigBytes
+    ///
+    /// - If stored in memory, returns a clone of the bytes.
+    /// - If stored on disk, returns the file path of the temporary file.
+    pub fn read(&self) -> IoResult<BigBytesReadResult> {
+        match self {
+            BigBytes::InMemory(vec) => Ok(BigBytesReadResult::InMemory(vec.clone())),
+            BigBytes::OnDisk(temp_file) => {
+                // Flush to be safe, then return the file path.
+                temp_file.as_file().sync_all()?;
+                Ok(BigBytesReadResult::OnDisk(temp_file.path().to_path_buf()))
+            }
+        }
+    }
+
+    /// Turn this into a value that can be used in Ruby.
+    pub fn as_value(&self) -> Option<Value> {
+        match self {
+            BigBytes::InMemory(bytes) => {
+                let bytes = Bytes::from(bytes.to_owned());
+                if bytes.is_empty() {
+                    None
+                } else {
+                    Some(bytes.into_value())
+                }
+            }
+            BigBytes::OnDisk(path) => {
+                let ruby = Ruby::get().unwrap();
+                let rarray = ruby.ary_new();
+                rarray.push(path.path().to_str().unwrap().into_value()).ok();
+                Some(rarray.into_value())
+            }
+        }
+    }
+}
+
+impl Drop for BigBytes {
+    fn drop(&mut self) {
+        match self {
+            BigBytes::InMemory(_) => {}
+            BigBytes::OnDisk(path) => {
+                let _ = std::fs::remove_file(path);
+            }
+        }
+    }
+}
+
+impl Default for BigBytes {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+impl Write for BigBytes {
+    fn write(&mut self, buf: &[u8]) -> IoResult<usize> {
+        match self {
+            BigBytes::InMemory(vec) => {
+                // Check if writing the new bytes would exceed the threshold.
+                if vec.len() + buf.len() > THRESHOLD {
+                    let mut tmp = NamedTempFile::new()?;
+                    tmp.write_all(vec)?;
+                    tmp.write_all(buf)?;
+                    *self = BigBytes::OnDisk(tmp);
+                    Ok(buf.len())
+                } else {
+                    vec.extend_from_slice(buf);
+                    Ok(buf.len())
+                }
+            }
+            BigBytes::OnDisk(tmp_file) => tmp_file.write(buf),
+        }
+    }
+
+    fn flush(&mut self) -> IoResult<()> {
+        match self {
+            BigBytes::InMemory(_) => Ok(()),
+            BigBytes::OnDisk(tmp_file) => tmp_file.flush(),
+        }
+    }
+}

data/ext/itsi_server/src/ruby_types/itsi_body_proxy/mod.rs

@@ -0,0 +1,143 @@
+pub mod big_bytes;
+use big_bytes::BigBytes;
+use bytes::Bytes;
+use futures::executor::block_on;
+use http_body_util::{BodyDataStream, BodyExt};
+use hyper::body::Incoming;
+use magnus::{error::Result as MagnusResult, scan_args, IntoValue, RString, Ruby, Value};
+use parking_lot::Mutex;
+use std::sync::{
+    atomic::{self, AtomicBool},
+    Arc,
+};
+use tokio_stream::StreamExt;
+
+use crate::server::size_limited_incoming::SizeLimitedIncoming;
+
+#[magnus::wrap(class = "Itsi::BodyProxy", free_immediately, size)]
+#[derive(Debug, Clone)]
+pub struct ItsiBodyProxy {
+    pub incoming: Arc<Mutex<BodyDataStream<SizeLimitedIncoming<Incoming>>>>,
+    pub closed: Arc<AtomicBool>,
+    pub buf: Arc<Mutex<Vec<u8>>>,
+}
+
+pub enum ItsiBody {
+    Buffered(BigBytes),
+    Stream(ItsiBodyProxy),
+}
+
+impl ItsiBody {
+    pub fn into_value(&self) -> Option<Value> {
+        match self {
+            ItsiBody::Buffered(bytes) => bytes.as_value(),
+            ItsiBody::Stream(proxy) => Some(proxy.clone().into_value()),
+        }
+    }
+}
+impl ItsiBodyProxy {
+    pub fn new(incoming: SizeLimitedIncoming<Incoming>) -> Self {
+        ItsiBodyProxy {
+            incoming: Arc::new(Mutex::new(incoming.into_data_stream())),
+            closed: Arc::new(AtomicBool::new(false)),
+            buf: Arc::new(Mutex::new(vec![])),
+        }
+    }
+    /// Read up to the next line-break OR EOF
+    pub fn gets(&self) -> MagnusResult<Option<Bytes>> {
+        self.verify_open()?;
+        let mut stream = self.incoming.lock();
+        let mut buf = self.buf.lock();
+        while !buf.contains(&b'\n') {
+            if let Some(chunk) = block_on(stream.next()) {
+                let chunk = chunk.map_err(|err| {
+                    magnus::Error::new(
+                        magnus::exception::standard_error(),
+                        format!("Error reading body {:?}", err),
+                    )
+                })?;
+                buf.extend_from_slice(&chunk);
+            } else {
+                break;
+            }
+        }
+        if let Some(pos) = buf.iter().position(|&x| x == b'\n') {
+            let line = buf.drain(..=pos).collect::<Vec<u8>>();
+            Ok(Some(line.into()))
+        } else if !buf.is_empty() {
+            let line = buf.drain(..).collect::<Vec<u8>>();
+            Ok(Some(line.into()))
+        } else {
+            Ok(None)
+        }
+    }
+
+    pub fn read(&self, args: &[Value]) -> MagnusResult<Option<RString>> {
+        self.verify_open()?;
+        let scanned =
+            scan_args::scan_args::<(), (Option<usize>, Option<RString>), (), (), (), ()>(args)?;
+        let (length, mut buffer) = scanned.optional;
+        let mut stream = self.incoming.lock();
+        let mut buf = self.buf.lock();
+
+        while length.is_none_or(|target_length| buf.len() < target_length) {
+            if let Some(chunk) = block_on(stream.next()) {
+                let chunk = chunk.map_err(|err| {
+                    magnus::Error::new(
+                        magnus::exception::standard_error(),
+                        format!("Error reading body {:?}", err),
+                    )
+                })?;
+                buf.extend_from_slice(&chunk);
+            } else if length.is_some() {
+                return Ok(None);
+            } else {
+                break;
+            }
+        }
+        let output_string = buffer.take().unwrap_or(RString::buf_new(buf.len()));
+        output_string.cat(buf.clone());
+        buf.clear();
+        Ok(Some(output_string))
+    }
+
+    pub fn to_bytes(&self) -> MagnusResult<Vec<u8>> {
+        self.verify_open()?;
+        let mut stream = self.incoming.lock();
+        let mut buf = self.buf.lock();
+
+        while let Some(chunk) = block_on(stream.next()) {
+            let chunk = chunk.map_err(|err| {
+                magnus::Error::new(
+                    magnus::exception::standard_error(),
+                    format!("Error reading body {:?}", err),
+                )
+            })?;
+            buf.extend_from_slice(&chunk);
+        }
+
+        Ok(buf.clone())
+    }
+
+    /// Equivalent to calling gets and yielding it, until we reach EOF
+    pub fn each(ruby: &Ruby, rbself: &Self) -> MagnusResult<()> {
+        let proc = ruby.block_proc()?;
+        while let Some(str) = rbself.gets()? {
+            proc.call::<_, Value>((str,))?;
+        }
+        Ok(())
+    }
+
+    fn verify_open(&self) -> MagnusResult<()> {
+        if self.closed.load(atomic::Ordering::SeqCst) {
+            return Err(magnus::Error::new(
+                magnus::exception::standard_error(),
+                "Body stream is closed",
+            ));
+        }
+        Ok(())
+    }
+    pub fn close(&self) {
+        self.closed.store(true, atomic::Ordering::SeqCst);
+    }
+}

data/ext/itsi_server/src/ruby_types/itsi_grpc_call.rs

@@ -0,0 +1,344 @@
+use super::itsi_grpc_response_stream::ItsiGrpcResponseStream;
+use crate::prelude::*;
+use crate::server::http_message_types::{HttpRequest, HttpResponse};
+use crate::server::{byte_frame::ByteFrame, request_job::RequestJob};
+use crate::services::itsi_http_service::HttpRequestContext;
+use async_compression::futures::bufread::{GzipDecoder, GzipEncoder, ZlibDecoder, ZlibEncoder};
+use bytes::Bytes;
+use derive_more::Debug;
+use futures::{executor::block_on, io::Cursor, AsyncReadExt};
+use http::{request::Parts, Response, StatusCode};
+use http_body_util::{combinators::BoxBody, BodyExt, Empty};
+use itsi_error::CLIENT_CONNECTION_CLOSED;
+use itsi_rb_helpers::{print_rb_backtrace, HeapValue};
+use itsi_tracing::debug;
+use magnus::{
+    block::Proc,
+    error::{ErrorType, Result as MagnusResult},
+    Error, Symbol,
+};
+use magnus::{
+    value::{LazyId, ReprValue},
+    Ruby, Value,
+};
+use regex::Regex;
+use std::sync::LazyLock;
+use std::{collections::HashMap, sync::Arc, time::Instant};
+use tokio::sync::mpsc::{self};
+
+static ID_MESSAGE: LazyId = LazyId::new("message");
+static MIN_GZIP_SIZE: u32 = 128;
+static MIN_DEFLATE_SIZE: u32 = 128;
+static METHOD_NAME_REGEX: LazyLock<Regex> =
+    LazyLock::new(|| Regex::new(r"([a-z])([A-Z])").expect("Failed to compile regex"));
+
+#[derive(Debug)]
+#[magnus::wrap(class = "Itsi::GrpcCall", free_immediately, size)]
+pub struct ItsiGrpcCall {
+    pub parts: Parts,
+    pub start: Instant,
+    pub compression_in: CompressionAlgorithm,
+    pub compression_out: CompressionAlgorithm,
+    #[debug(skip)]
+    pub context: HttpRequestContext,
+    #[debug(skip)]
+    pub stream: ItsiGrpcResponseStream,
+}
+
+#[derive(Debug, Clone)]
+pub enum CompressionAlgorithm {
+    None,
+    Deflate,
+    Gzip,
+}
+
+impl ItsiGrpcCall {
+    pub fn service_name(&self) -> MagnusResult<String> {
+        let path = self.parts.uri.path();
+        Ok(path.split('/').nth_back(1).unwrap().to_string())
+    }
+
+    pub fn method_name(&self) -> MagnusResult<Symbol> {
+        let path = self.parts.uri.path();
+        let method_name = path.split('/').nth_back(0).unwrap();
+        let snake_case_method_name = METHOD_NAME_REGEX
+            .replace_all(method_name, "${1}_${2}")
+            .to_lowercase();
+        Ok(Symbol::new(snake_case_method_name))
+    }
+
+    pub fn stream(&self) -> MagnusResult<ItsiGrpcResponseStream> {
+        Ok(self.stream.clone())
+    }
+
+    pub fn timeout(&self) -> MagnusResult<Option<f64>> {
+        let timeout_str = self
+            .parts
+            .headers
+            .get("grpc-timeout")
+            .and_then(|hv| hv.to_str().ok())
+            .unwrap_or("");
+        Ok(parse_grpc_timeout(timeout_str).ok())
+    }
+
+    pub fn is_cancelled(&self) -> MagnusResult<bool> {
+        self.stream.is_cancelled()
+    }
+
+    pub fn add_headers(&self, headers: HashMap<Bytes, Vec<Bytes>>) -> MagnusResult<()> {
+        self.stream.add_headers(headers)
+    }
+
+    pub fn content_type_str(&self) -> &str {
+        self.parts
+            .headers
+            .get("Content-Type")
+            .and_then(|hv| hv.to_str().ok())
+            .unwrap_or("application/x-www-form-urlencoded")
+    }
+
+    pub fn is_json(&self) -> bool {
+        self.content_type_str() == "application/json"
+    }
+
+    pub fn process(self, ruby: &Ruby, app_proc: Arc<HeapValue<Proc>>) -> magnus::error::Result<()> {
+        let response = self.stream.clone();
+        let result = app_proc.call::<_, Value>((self,));
+        if let Err(err) = result {
+            Self::internal_error(ruby, response, err);
+        }
+        Ok(())
+    }
+
+    pub fn internal_error(_ruby: &Ruby, stream: ItsiGrpcResponseStream, err: Error) {
+        if let Some(rb_err) = err.value() {
+            print_rb_backtrace(rb_err);
+            stream.internal_server_error(err.to_string());
+        } else {
+            stream.internal_server_error(err.to_string());
+        }
+    }
+
+    pub(crate) async fn process_request(
+        app: Arc<HeapValue<Proc>>,
+        hyper_request: HttpRequest,
+        context: &HttpRequestContext,
+        nonblocking: bool,
+    ) -> itsi_error::Result<HttpResponse> {
+        let (request, mut receiver) = ItsiGrpcCall::new(hyper_request, context).await;
+        let shutdown_channel = context.service.shutdown_channel.clone();
+        let response_stream = request.stream.clone();
+        let sender = if nonblocking {
+            &context.nonblocking_sender
+        } else {
+            &context.sender
+        };
+        match sender
+            .send(RequestJob::ProcessGrpcRequest(request, app))
+            .await
+        {
+            Err(err) => {
+                error!("Error occurred: {}", err);
+                let mut response = Response::new(BoxBody::new(Empty::new()));
+                *response.status_mut() = StatusCode::BAD_REQUEST;
+                Ok(response)
+            }
+            _ => match receiver.recv().await {
+                Some(first_frame) => Ok(response_stream
+                    .build_response(first_frame, receiver, shutdown_channel)
+                    .await),
+                None => Ok(Response::new(BoxBody::new(Empty::new()))),
+            },
+        }
+    }
+
+    pub fn is_connection_closed_err(ruby: &Ruby, err: &Error) -> bool {
+        match err.error_type() {
+            ErrorType::Jump(_) => false,
+            ErrorType::Error(_, _) => false,
+            ErrorType::Exception(exception) => {
+                exception.is_kind_of(ruby.exception_eof_error())
+                    && err
+                        .value()
+                        .map(|v| {
+                            v.funcall::<_, _, String>(*ID_MESSAGE, ())
+                                .unwrap_or("".to_string())
+                                .eq(CLIENT_CONNECTION_CLOSED)
+                        })
+                        .unwrap_or(false)
+            }
+        }
+    }
+
+    pub fn should_compress_output(&self, message_size: u32) -> bool {
+        match self.compression_out {
+            CompressionAlgorithm::Gzip => message_size > MIN_GZIP_SIZE,
+            CompressionAlgorithm::Deflate => message_size > MIN_DEFLATE_SIZE,
+            CompressionAlgorithm::None => false,
+        }
+    }
+
+    pub fn compress_output(&self, bytes: Bytes) -> MagnusResult<Bytes> {
+        match self.compression_out {
+            CompressionAlgorithm::Gzip => Self::compress_gzip(bytes),
+            CompressionAlgorithm::Deflate => Self::compress_deflate(bytes),
+            CompressionAlgorithm::None => Ok(bytes),
+        }
+    }
+
+    pub fn decompress_input(&self, bytes: Bytes) -> MagnusResult<Bytes> {
+        match self.compression_in {
+            CompressionAlgorithm::Gzip => Self::decompress_gzip(bytes),
+            CompressionAlgorithm::Deflate => Self::decompress_deflate(bytes),
+            CompressionAlgorithm::None => Ok(bytes),
+        }
+    }
+
+    fn decompress_deflate(input: Bytes) -> MagnusResult<Bytes> {
+        let cursor = Cursor::new(input);
+        let mut decoder = ZlibDecoder::new(cursor);
+
+        let result = block_on(async {
+            let mut output = Vec::new();
+            decoder.read_to_end(&mut output).await?;
+            Ok(Bytes::from(output))
+        })
+        .map_err(|e: std::io::Error| {
+            Error::new(
+                magnus::exception::standard_error(),
+                format!("deflate decompression failed: {}", e),
+            )
+        })?;
+
+        Ok(result)
+    }
+
+    fn decompress_gzip(input: Bytes) -> MagnusResult<Bytes> {
+        let cursor = Cursor::new(input);
+        let mut decoder = GzipDecoder::new(cursor);
+
+        let result = block_on(async {
+            let mut output = Vec::new();
+            decoder.read_to_end(&mut output).await?;
+            Ok(Bytes::from(output))
+        })
+        .map_err(|e: std::io::Error| {
+            Error::new(
+                magnus::exception::standard_error(),
+                format!("gzip decompression failed: {}", e),
+            )
+        })?;
+
+        Ok(result)
+    }
+
+    fn compress_gzip(input: Bytes) -> MagnusResult<Bytes> {
+        let mut output = Vec::with_capacity(input.len() / 2);
+        let cursor = Cursor::new(input);
+        let mut encoder = GzipEncoder::new(cursor);
+
+        let result = block_on(async {
+            encoder.read_to_end(&mut output).await?;
+            Ok::<Bytes, std::io::Error>(output.into())
+        })
+        .map_err(|e| {
+            Error::new(
+                magnus::exception::standard_error(),
+                format!("gzip compression failed: {e}"),
+            )
+        })?;
+
+        Ok(result)
+    }
+
+    fn compress_deflate(input: Bytes) -> MagnusResult<Bytes> {
+        let mut output = Vec::with_capacity(input.len() / 2);
+        let cursor = Cursor::new(input);
+        let mut encoder = ZlibEncoder::new(cursor);
+
+        let result = block_on(async {
+            encoder.read_to_end(&mut output).await?;
+            Ok::<Bytes, std::io::Error>(output.into())
+        })
+        .map_err(|e| {
+            Error::new(
+                magnus::exception::standard_error(),
+                format!("deflate compression failed: {e}"),
+            )
+        })?;
+
+        Ok(result)
+    }
+
+    pub(crate) async fn new(
+        request: HttpRequest,
+        context: &HttpRequestContext,
+    ) -> (ItsiGrpcCall, mpsc::Receiver<ByteFrame>) {
+        let (parts, body) = request.into_parts();
+        let response_channel = mpsc::channel::<ByteFrame>(100);
+        let compression_in: CompressionAlgorithm = match parts.headers.get("grpc-encoding") {
+            Some(encoding) => match encoding.to_str() {
+                Ok(encoding) => match encoding {
+                    "gzip" => CompressionAlgorithm::Gzip,
+                    "deflate" => CompressionAlgorithm::Deflate,
+                    _ => CompressionAlgorithm::None,
+                },
+                Err(_) => CompressionAlgorithm::None,
+            },
+            None => CompressionAlgorithm::None,
+        };
+        let compression_out: CompressionAlgorithm = match parts.headers.get("grpc-accept-encoding")
+        {
+            Some(accept_encoding) => match accept_encoding.to_str() {
+                Ok(accept_encoding) => {
+                    let encodings: Vec<&str> =
+                        accept_encoding.split(',').map(|s| s.trim()).collect();
+                    if encodings.contains(&"gzip") {
+                        CompressionAlgorithm::Gzip
+                    } else if encodings.contains(&"deflate") {
+                        CompressionAlgorithm::Deflate
+                    } else {
+                        CompressionAlgorithm::None
+                    }
+                }
+                Err(_) => CompressionAlgorithm::None,
+            },
+            None => CompressionAlgorithm::None,
+        };
+        (
+            Self {
+                context: context.clone(),
+                start: Instant::now(),
+                compression_out: compression_out.clone(),
+                compression_in,
+                parts,
+                stream: ItsiGrpcResponseStream::new(
+                    compression_out,
+                    response_channel.0,
+                    body.into_data_stream(),
+                )
+                .await,
+            },
+            response_channel.1,
+        )
+    }
+}
+
+fn parse_grpc_timeout(timeout_str: &str) -> Result<f64> {
+    if timeout_str.len() < 2 {
+        return Err("Timeout string too short".into());
+    }
+    let (value_str, unit) = timeout_str.split_at(timeout_str.len() - 1);
+    let value: u64 = value_str.parse().map_err(|_| "Invalid timeout value")?;
+    let duration_secs = match unit {
+        "n" => value as f64 / 1_000_000_000.0, // nanoseconds
+        "u" => value as f64 / 1_000_000.0,     // microseconds
+        "m" => value as f64 / 1_000.0,         // milliseconds
+        "S" => value as f64,                   // seconds
+        "M" => value as f64 * 60.0,            // minutes
+        "H" => value as f64 * 3600.0,          // hours
+        _ => return Err("Invalid timeout unit".into()),
+    };
+
+    Ok(duration_secs)
+}