vinted-prometheus-client-mmap 1.5.0-x86_64-linux

data/ext/fast_mmaped_file_rs/src/file_info.rs

@@ -0,0 +1,240 @@
+use magnus::exception::*;
+use magnus::{Error, RString, Symbol, Value};
+use std::ffi::OsString;
+use std::fs::File;
+use std::io::{self, Read, Seek};
+use std::os::unix::ffi::OsStringExt;
+use std::path::PathBuf;
+
+use crate::err;
+use crate::error::{MmapError, RubyError};
+use crate::util;
+use crate::Result;
+
+/// The details of a `*.db` file.
+#[derive(Debug)]
+pub struct FileInfo {
+    pub file: File,
+    pub path: PathBuf,
+    pub len: usize,
+    pub multiprocess_mode: Symbol,
+    pub type_: Symbol,
+    pub pid: String,
+}
+
+impl FileInfo {
+    /// Receive the details of a file from Ruby and store as a `FileInfo`.
+    pub fn open_from_params(params: &[Value; 4]) -> magnus::error::Result<Self> {
+        if params.len() != 4 {
+            return Err(err!(
+                arg_error(),
+                "wrong number of arguments {} instead of 4",
+                params.len()
+            ));
+        }
+
+        let filepath = RString::from_value(params[0])
+            .ok_or_else(|| err!(arg_error(), "can't convert filepath to String"))?;
+
+        // SAFETY: We immediately copy the string buffer from Ruby, preventing
+        // it from being mutated out from under us.
+        let path_bytes: Vec<_> = unsafe { filepath.as_slice().to_owned() };
+        let path = PathBuf::from(OsString::from_vec(path_bytes));
+
+        let mut file = File::open(&path).map_err(|_| {
+            err!(
+                arg_error(),
+                "Can't open {}, errno: {}",
+                path.display(),
+                util::errno()
+            )
+        })?;
+
+        let stat = file
+            .metadata()
+            .map_err(|_| err!(io_error(), "Can't stat file, errno: {}", util::errno()))?;
+
+        let length = util::cast_chk::<_, usize>(stat.len(), "file size")?;
+
+        let multiprocess_mode = Symbol::from_value(params[1])
+            .ok_or_else(|| err!(arg_error(), "expected multiprocess_mode to be a symbol"))?;
+
+        let type_ = Symbol::from_value(params[2])
+            .ok_or_else(|| err!(arg_error(), "expected file type to be a symbol"))?;
+
+        let pid = RString::from_value(params[3])
+            .ok_or_else(|| err!(arg_error(), "expected pid to be a String"))?;
+
+        file.rewind()
+            .map_err(|_| err!(io_error(), "Can't fseek 0, errno: {}", util::errno()))?;
+
+        Ok(Self {
+            file,
+            path,
+            len: length,
+            multiprocess_mode,
+            type_,
+            pid: pid.to_string()?,
+        })
+    }
+
+    /// Read the contents of the associated file into the buffer provided by
+    /// the caller.
+    pub fn read_from_file(&mut self, buf: &mut Vec<u8>) -> Result<()> {
+        buf.clear();
+        buf.try_reserve(self.len).map_err(|_| {
+            MmapError::legacy(
+                format!("Can't malloc {}, errno: {}", self.len, util::errno()),
+                RubyError::Io,
+            )
+        })?;
+
+        match self.file.read_to_end(buf) {
+            Ok(n) if n == self.len => Ok(()),
+            // A worker may expand the file between our `stat` and `read`, no harm done.
+            Ok(n) if n > self.len => {
+                self.len = n;
+                Ok(())
+            }
+            Ok(_) => Err(MmapError::io(
+                "read",
+                &self.path,
+                io::Error::from(io::ErrorKind::UnexpectedEof),
+            )),
+            Err(e) => Err(MmapError::io("read", &self.path, e)),
+        }
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use magnus::{eval, RArray, Symbol};
+    use rand::{thread_rng, Rng};
+    use sha2::{Digest, Sha256};
+    use std::fs;
+    use std::io::Write;
+
+    use super::*;
+    use crate::testhelper::TestFile;
+
+    #[test]
+    fn test_open_from_params() {
+        let _cleanup = unsafe { magnus::embed::init() };
+        let ruby = magnus::Ruby::get().unwrap();
+        crate::init(&ruby).unwrap();
+
+        let file_data = b"foobar";
+        let TestFile {
+            file: _file,
+            path,
+            dir: _dir,
+        } = TestFile::new(file_data);
+
+        let pid = "worker-1_0";
+        let args = RArray::from_value(
+            eval(&format!("['{}', :max, :gauge, '{pid}']", path.display())).unwrap(),
+        )
+        .unwrap();
+        let arg0 = args.shift().unwrap();
+        let arg1 = args.shift().unwrap();
+        let arg2 = args.shift().unwrap();
+        let arg3 = args.shift().unwrap();
+
+        let out = FileInfo::open_from_params(&[arg0, arg1, arg2, arg3]);
+        assert!(out.is_ok());
+
+        let out = out.unwrap();
+
+        assert_eq!(out.path, path);
+        assert_eq!(out.len, file_data.len());
+        assert_eq!(out.multiprocess_mode, Symbol::new("max"));
+        assert_eq!(out.type_, Symbol::new("gauge"));
+        assert_eq!(out.pid, pid);
+    }
+
+    #[test]
+    fn test_read_from_file() {
+        let _cleanup = unsafe { magnus::embed::init() };
+        let ruby = magnus::Ruby::get().unwrap();
+        crate::init(&ruby).unwrap();
+
+        const BUF_LEN: usize = 1 << 20; // 1MiB
+
+        // Create a buffer with random data.
+        let mut buf = vec![0u8; BUF_LEN];
+        thread_rng().fill(buf.as_mut_slice());
+
+        let TestFile {
+            file,
+            path,
+            dir: _dir,
+        } = TestFile::new(&buf);
+
+        let mut info = FileInfo {
+            file,
+            path: path.clone(),
+            len: buf.len(),
+            multiprocess_mode: Symbol::new("puma"),
+            type_: Symbol::new("max"),
+            pid: "worker-0_0".to_string(),
+        };
+
+        let mut out_buf = Vec::new();
+        info.read_from_file(&mut out_buf).unwrap();
+
+        assert_eq!(buf.len(), out_buf.len(), "buffer lens");
+
+        let mut in_hasher = Sha256::new();
+        in_hasher.update(&buf);
+        let in_hash = in_hasher.finalize();
+
+        let mut out_hasher = Sha256::new();
+        out_hasher.update(&out_buf);
+        let out_hash = out_hasher.finalize();
+
+        assert_eq!(in_hash, out_hash, "content hashes");
+    }
+
+    #[test]
+    fn test_read_from_file_resized() {
+        let _cleanup = unsafe { magnus::embed::init() };
+        let ruby = magnus::Ruby::get().unwrap();
+        crate::init(&ruby).unwrap();
+
+        const BUF_LEN: usize = 1 << 14; // 16KiB
+
+        // Create a buffer with random data.
+        let mut buf = vec![0u8; BUF_LEN];
+        thread_rng().fill(buf.as_mut_slice());
+
+        let TestFile {
+            file,
+            path,
+            dir: _dir,
+        } = TestFile::new(&buf);
+
+        let mut info = FileInfo {
+            file,
+            path: path.clone(),
+            len: buf.len(),
+            multiprocess_mode: Symbol::new("puma"),
+            type_: Symbol::new("max"),
+            pid: "worker-0_0".to_string(),
+        };
+
+        let mut resized_file = fs::OpenOptions::new()
+            .write(true)
+            .append(true)
+            .open(path)
+            .unwrap();
+
+        // Write data to file after it has been `stat`ed in the
+        // constructor.
+        resized_file.write_all(&[1; 1024]).unwrap();
+
+        let mut out_buf = Vec::new();
+        info.read_from_file(&mut out_buf).unwrap();
+
+        assert_eq!(BUF_LEN + 1024, info.len, "resized file updated len");
+    }
+}

data/ext/fast_mmaped_file_rs/src/lib.rs

@@ -0,0 +1,89 @@
+use magnus::exception::*;
+use magnus::prelude::*;
+use magnus::value::{Fixnum, Lazy, LazyId};
+use magnus::{class, define_class, exception, function, method, Ruby};
+use std::mem::size_of;
+
+use crate::mmap::MmapedFile;
+
+pub mod error;
+pub mod file_entry;
+pub mod file_info;
+mod macros;
+pub mod map;
+pub mod mmap;
+pub mod raw_entry;
+pub mod util;
+pub mod exemplars;
+
+pub mod io {
+    pub mod prometheus {
+        pub mod client {
+            include!(concat!(env!("OUT_DIR"), "/io.prometheus.client.rs"));
+        }
+    }
+}
+
+#[cfg(test)]
+mod testhelper;
+
+type Result<T> = std::result::Result<T, crate::error::MmapError>;
+
+const MAP_SHARED: i64 = libc::MAP_SHARED as i64;
+const HEADER_SIZE: usize = 2 * size_of::<u32>();
+
+static SYM_GAUGE: LazyId = LazyId::new("gauge");
+static SYM_MIN: LazyId = LazyId::new("min");
+static SYM_MAX: LazyId = LazyId::new("max");
+static SYM_LIVESUM: LazyId = LazyId::new("livesum");
+static SYM_PID: LazyId = LazyId::new("pid");
+static SYM_SAMPLES: LazyId = LazyId::new("samples");
+
+static PROM_EPARSING_ERROR: Lazy<ExceptionClass> = Lazy::new(|_| {
+    let prom_err = define_class(
+        "PrometheusParsingError",
+        exception::runtime_error().as_r_class(),
+    )
+    .expect("failed to create class `PrometheusParsingError`");
+    ExceptionClass::from_value(prom_err.as_value())
+        .expect("failed to create exception class from `PrometheusParsingError`")
+});
+
+#[magnus::init]
+fn init(ruby: &Ruby) -> magnus::error::Result<()> {
+    // Initialize the static symbols
+    LazyId::force(&SYM_GAUGE, ruby);
+    LazyId::force(&SYM_MIN, ruby);
+    LazyId::force(&SYM_MAX, ruby);
+    LazyId::force(&SYM_LIVESUM, ruby);
+    LazyId::force(&SYM_PID, ruby);
+    LazyId::force(&SYM_SAMPLES, ruby);
+
+    // Initialize `PrometheusParsingError` class.
+    Lazy::force(&PROM_EPARSING_ERROR, ruby);
+
+    let klass = define_class("FastMmapedFileRs", class::object())?;
+    klass.undef_default_alloc_func();
+
+    // UNWRAP: We know `MAP_SHARED` fits in a `Fixnum`.
+    klass.const_set("MAP_SHARED", Fixnum::from_i64(MAP_SHARED).unwrap())?;
+
+    klass.define_singleton_method("to_metrics", function!(MmapedFile::to_metrics, 1))?;
+    klass.define_singleton_method("to_protobuf", function!(MmapedFile::to_protobuf, 1))?;
+
+    // Required for subclassing to work
+    klass.define_alloc_func::<MmapedFile>();
+    klass.define_singleton_method("new", method!(MmapedFile::new, -1))?;
+    klass.define_method("initialize", method!(MmapedFile::initialize, 1))?;
+    klass.define_method("slice", method!(MmapedFile::slice, -1))?;
+    klass.define_method("sync", method!(MmapedFile::sync, -1))?;
+    klass.define_method("munmap", method!(MmapedFile::munmap, 0))?;
+
+    klass.define_method("used", method!(MmapedFile::load_used, 0))?;
+    klass.define_method("used=", method!(MmapedFile::save_used, 1))?;
+    klass.define_method("fetch_entry", method!(MmapedFile::fetch_entry, 3))?;
+    klass.define_method("upsert_entry", method!(MmapedFile::upsert_entry, 3))?;
+    klass.define_method("upsert_exemplar", method!(MmapedFile::upsert_exemplar, 5))?;
+
+    Ok(())
+}

data/ext/fast_mmaped_file_rs/src/macros.rs

@@ -0,0 +1,14 @@
+#[macro_export]
+macro_rules! err {
+    (with_errno: $err_t:expr, $($arg:expr),*) => {
+        {
+            let err = format!($($arg),*);
+            let strerror = strerror(errno());
+            Error::new($err_t, format!("{err} ({strerror})"))
+        }
+    };
+
+    ($err_t:expr, $($arg:expr),*) => {
+        Error::new($err_t, format!($($arg),*))
+    };
+}