zellij/zellij-server/src/logging_pipe.rs
bjorn3 7d7848cddc
dependencies: switch from Wasmer to Wasmtime (#3349)
* Remove ForeignFunctionEnv wrapper around PluginEnv

This will enable PluginEnv to be the Store context when migrating to
Wasmtime.

* Pass PluginEnv by value to load_plugin_instance

This will allow removing the Clone impl from PluginEnv when migrating to
Wasmtime as required by the missing Clone impl on Wasmtime's WasiCtx.

* Avoid passing a Store around when an Engine is enough

* Pass PluginEnv to the wasi read/write functions

Wasmtime requires storing the read/write end of the pipe outside of the
WasiCtx. Passing PluginEnv to these functions allows storing them in the
PluginEnv.

* Migrate to Wasmtime

* Switch from wasi-common to wasmtime-wasi

* Reduce verbosity of wasmtime_wasi logs

* Increase startup delay

To wait for all plugins to be compiled.

* Disable some wasmtime features

* Update to Wasmtime 21.0.1
2024-06-28 16:47:43 +02:00

206 lines
6.3 KiB
Rust

use std::{collections::VecDeque, io::Write};
use crate::plugins::PluginId;
use log::{debug, error};
use zellij_utils::{errors::prelude::*, serde};
use chrono::prelude::*;
use serde::{Deserialize, Serialize};
// 16kB log buffer
const ZELLIJ_MAX_PIPE_BUFFER_SIZE: usize = 16_384;
#[derive(Debug, Serialize, Deserialize)]
#[serde(crate = "self::serde")]
pub struct LoggingPipe {
buffer: VecDeque<u8>,
plugin_name: String,
plugin_id: PluginId,
}
impl LoggingPipe {
pub fn new(plugin_name: &str, plugin_id: PluginId) -> LoggingPipe {
LoggingPipe {
buffer: VecDeque::new(),
plugin_name: String::from(plugin_name),
plugin_id,
}
}
fn log_message(&self, message: &str) {
debug!(
"|{:<25.25}| {} [{:<10.15}] {}",
self.plugin_name,
Local::now().format("%Y-%m-%d %H:%M:%S.%3f"),
format!("id: {}", self.plugin_id),
message
);
}
}
impl Write for LoggingPipe {
fn write(&mut self, buf: &[u8]) -> std::io::Result<usize> {
if self.buffer.len() + buf.len() > ZELLIJ_MAX_PIPE_BUFFER_SIZE {
let error_msg =
"Exceeded log buffer size. Make sure that your plugin calls flush on stderr on \
valid UTF-8 symbol boundary. Additionally, make sure that your log message contains \
endline \\n symbol.";
error!("{}: {}", self.plugin_name, error_msg);
self.buffer.clear();
return Err(std::io::Error::new(
std::io::ErrorKind::InvalidData,
error_msg,
));
}
self.buffer.extend(buf);
Ok(buf.len())
}
// When we flush, check if current buffer is valid utf8 string, split by '\n' and truncate buffer in the process.
// We assume that eventually, flush will be called on valid string boundary (i.e. std::str::from_utf8(..).is_ok() returns true at some point).
// Above assumption might not be true, in which case we'll have to think about it. Make it simple for now.
fn flush(&mut self) -> std::io::Result<()> {
self.buffer.make_contiguous();
match std::str::from_utf8(self.buffer.as_slices().0) {
Ok(converted_buffer) => {
if converted_buffer.contains('\n') {
let mut consumed_bytes = 0;
let mut split_converted_buffer = converted_buffer.split('\n').peekable();
while let Some(msg) = split_converted_buffer.next() {
if split_converted_buffer.peek().is_none() {
// Log last chunk iff the last char is endline. Otherwise do not do it.
if converted_buffer.ends_with('\n') && !msg.is_empty() {
self.log_message(msg);
consumed_bytes += msg.len() + 1;
}
} else {
self.log_message(msg);
consumed_bytes += msg.len() + 1;
}
}
drop(self.buffer.drain(..consumed_bytes));
}
},
Err(e) => Err::<(), _>(e)
.context("failed to flush logging pipe buffer")
.non_fatal(),
}
Ok(())
}
}
// Unit tests
#[cfg(test)]
mod logging_pipe_test {
use super::*;
#[test]
fn write_without_endl_does_not_consume_buffer_after_flush() {
let mut pipe = LoggingPipe::new("TestPipe", 0);
let test_buffer = b"Testing write";
pipe.write_all(test_buffer).expect("Err write");
pipe.flush().expect("Err flush");
assert_eq!(pipe.buffer.len(), test_buffer.len());
}
#[test]
fn write_with_single_endl_at_the_end_consumes_whole_buffer_after_flush() {
let mut pipe = LoggingPipe::new("TestPipe", 0);
let test_buffer = b"Testing write \n";
pipe.write_all(test_buffer).expect("Err write");
pipe.flush().expect("Err flush");
assert_eq!(pipe.buffer.len(), 0);
}
#[test]
fn write_with_endl_in_the_middle_consumes_buffer_up_to_endl_after_flush() {
let mut pipe = LoggingPipe::new("TestPipe", 0);
let test_buffer = b"Testing write \n";
let test_buffer2: &[_] = b"And the rest";
pipe.write_all(
[
test_buffer,
test_buffer,
test_buffer,
test_buffer,
test_buffer2,
]
.concat()
.as_slice(),
)
.expect("Err write");
pipe.flush().expect("Err flush");
assert_eq!(pipe.buffer.len(), test_buffer2.len());
}
#[test]
fn write_with_many_endl_consumes_whole_buffer_after_flush() {
let mut pipe = LoggingPipe::new("TestPipe", 0);
let test_buffer: &[_] = b"Testing write \n";
pipe.write_all(
[
test_buffer,
test_buffer,
test_buffer,
test_buffer,
test_buffer,
]
.concat()
.as_slice(),
)
.expect("Err write");
pipe.flush().expect("Err flush");
assert_eq!(pipe.buffer.len(), 0);
}
#[test]
fn write_with_incorrect_byte_boundary_does_not_crash() {
let mut pipe = LoggingPipe::new("TestPipe", 0);
let test_buffer = "😱".as_bytes();
// make sure it's not valid utf-8 string if we drop last symbol
assert!(std::str::from_utf8(&test_buffer[..test_buffer.len() - 1]).is_err());
pipe.write_all(&test_buffer[..test_buffer.len() - 1])
.expect("Err write");
pipe.flush().expect("Err flush");
assert_eq!(pipe.buffer.len(), test_buffer.len() - 1);
println!("len: {}, buf: {:?}", test_buffer.len(), test_buffer);
}
#[test]
fn write_with_many_endls_consumes_everything_after_flush() {
let mut pipe = LoggingPipe::new("TestPipe", 0);
let test_buffer: &[_] = b"Testing write \n";
pipe.write_all(
[test_buffer, test_buffer, b"\n", b"\n", b"\n"]
.concat()
.as_slice(),
)
.expect("Err write");
pipe.flush().expect("Err flush");
assert_eq!(pipe.buffer.len(), 0);
}
}