1 files changed, 132 insertions, 0 deletions

diff --git a/crates/proc_macro_api/src/version.rs b/crates/proc_macro_api/src/version.rs
new file mode 100644
index 000000000..11a7fb59a
--- /dev/null
+++ b/crates/proc_macro_api/src/version.rs
@@ -0,0 +1,132 @@
+//! Reading proc-macro rustc version information from binary data
+
+use std::{
+    fs::File,
+    io::{self, Read},
+    path::Path,
+};
+
+use memmap::Mmap;
+use object::read::{File as BinaryFile, Object, ObjectSection};
+use snap::read::FrameDecoder as SnapDecoder;
+
+#[derive(Debug)]
+pub(crate) struct RustCInfo {
+    pub(crate) version: (usize, usize, usize),
+    pub(crate) channel: String,
+    pub(crate) commit: String,
+    pub(crate) date: String,
+}
+
+pub(crate) fn read_info(dylib_path: &Path) -> io::Result<RustCInfo> {
+    macro_rules! err {
+        ($e:literal) => {
+            io::Error::new(io::ErrorKind::InvalidData, $e)
+        };
+    }
+
+    let ver_str = read_version(dylib_path)?;
+    let mut items = ver_str.split_whitespace();
+    let tag = items.next().ok_or(err!("version format error"))?;
+    if tag != "rustc" {
+        return Err(err!("version format error (No rustc tag)"));
+    }
+
+    let version_part = items.next().ok_or(err!("no version string"))?;
+    let mut version_parts = version_part.split("-");
+    let version = version_parts.next().ok_or(err!("no version"))?;
+    let channel = version_parts.next().unwrap_or_default().to_string();
+
+    let commit = items.next().ok_or(err!("no commit info"))?;
+    // remove (
+    if commit.len() == 0 {
+        return Err(err!("commit format error"));
+    }
+    let commit = commit[1..].to_string();
+    let date = items.next().ok_or(err!("no date info"))?;
+    // remove )
+    if date.len() == 0 {
+        return Err(err!("date format error"));
+    }
+    let date = date[0..date.len() - 2].to_string();
+
+    let version_numbers = version
+        .split(".")
+        .map(|it| it.parse::<usize>())
+        .collect::<Result<Vec<_>, _>>()
+        .map_err(|_| err!("version number error"))?;
+
+    if version_numbers.len() != 3 {
+        return Err(err!("version number format error"));
+    }
+    let version = (version_numbers[0], version_numbers[1], version_numbers[2]);
+
+    Ok(RustCInfo { version, channel, commit, date })
+}
+
+/// This is used inside read_version() to locate the ".rustc" section
+/// from a proc macro crate's binary file.
+fn read_section<'a>(dylib_binary: &'a [u8], section_name: &str) -> io::Result<&'a [u8]> {
+    BinaryFile::parse(dylib_binary)
+        .map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))?
+        .section_by_name(section_name)
+        .ok_or_else(|| io::Error::new(io::ErrorKind::InvalidData, "section read error"))?
+        .data()
+        .map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))
+}
+
+/// Check the version of rustc that was used to compile a proc macro crate's
+///
+/// binary file.
+/// A proc macro crate binary's ".rustc" section has following byte layout:
+/// * [b'r',b'u',b's',b't',0,0,0,5] is the first 8 bytes
+/// * ff060000 734e6150 is followed, it's the snappy format magic bytes,
+///   means bytes from here(including this sequence) are compressed in
+///   snappy compression format. Version info is inside here, so decompress
+///   this.
+/// The bytes you get after decompressing the snappy format portion has
+/// following layout:
+/// * [b'r',b'u',b's',b't',0,0,0,5] is the first 8 bytes(again)
+/// * [crate root bytes] next 4 bytes is to store crate root position,
+///   according to rustc's source code comment
+/// * [length byte] next 1 byte tells us how many bytes we should read next
+///   for the version string's utf8 bytes
+/// * [version string bytes encoded in utf8] <- GET THIS BOI
+/// * [some more bytes that we don really care but still there] :-)
+/// Check this issue for more about the bytes layout:
+/// https://github.com/rust-analyzer/rust-analyzer/issues/6174
+fn read_version(dylib_path: &Path) -> io::Result<String> {
+    let dylib_file = File::open(dylib_path)?;
+    let dylib_mmaped = unsafe { Mmap::map(&dylib_file) }?;
+
+    let dot_rustc = read_section(&dylib_mmaped, ".rustc")?;
+
+    let header = &dot_rustc[..8];
+    const EXPECTED_HEADER: [u8; 8] = [b'r', b'u', b's', b't', 0, 0, 0, 5];
+    // check if header is valid
+    if header != EXPECTED_HEADER {
+        return Err(io::Error::new(
+            io::ErrorKind::InvalidData,
+            format!("only metadata version 5 is supported, section header was: {:?}", header),
+        ));
+    }
+
+    let snappy_portion = &dot_rustc[8..];
+
+    let mut snappy_decoder = SnapDecoder::new(snappy_portion);
+
+    // the bytes before version string bytes, so this basically is:
+    // 8 bytes for [b'r',b'u',b's',b't',0,0,0,5]
+    // 4 bytes for [crate root bytes]
+    // 1 byte for length of version string
+    // so 13 bytes in total, and we should check the 13th byte
+    // to know the length
+    let mut bytes_before_version = [0u8; 13];
+    snappy_decoder.read_exact(&mut bytes_before_version)?;
+    let length = bytes_before_version[12]; // what? can't use -1 indexing?
+
+    let mut version_string_utf8 = vec![0u8; length as usize];
+    snappy_decoder.read_exact(&mut version_string_utf8)?;
+    let version_string = String::from_utf8(version_string_utf8);
+    version_string.map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))
+}