opaque-types 0.1.0

#![doc = include_str!("../README.md")]
#![forbid(unsafe_code)]

use std::path::{Path, PathBuf};

use anyhow::{anyhow, bail, Context, Result};

/// One source type and the opaque struct identifier generated for it.
#[derive(Clone, Debug)]
pub struct OpaqueType {
    /// Rust type expression as visible from the probe crate.
    pub rust_path: String,
    /// Emitted opaque struct identifier.
    pub opaque_name: String,
}

impl OpaqueType {
    pub fn new(rust_path: impl Into<String>, opaque_name: impl Into<String>) -> Self {
        Self {
            rust_path: rust_path.into(),
            opaque_name: opaque_name.into(),
        }
    }
}

/// Builder for generating layout-compatible opaque structs.
///
/// See the [crate-level documentation](crate) for exact parameter formats and
/// workspace setup.
#[derive(Clone, Debug)]
pub struct OpaqueTypes {
    source_manifest_dir: PathBuf,
    features: Vec<String>,
    no_default_features: bool,
    types: Vec<OpaqueType>,
    cargo_lock: Option<PathBuf>,
    build_dir: Option<PathBuf>,
}

impl OpaqueTypes {
    /// Creates a generator for the package whose `Cargo.toml` is inside
    /// `source_manifest_dir`.
    pub fn new(source_manifest_dir: impl Into<PathBuf>) -> Self {
        let build_dir = std::env::var_os("OUT_DIR").map(|o| PathBuf::from(o).join("opaque_probe"));
        Self {
            source_manifest_dir: source_manifest_dir.into(),
            features: Vec::new(),
            no_default_features: false,
            types: Vec::new(),
            cargo_lock: None,
            build_dir,
        }
    }

    /// Sets explicit, unqualified Cargo feature names for the source dependency.
    ///
    /// Each item must match a key in the source package's `[features]` table.
    /// This setting is independent of [`Self::default_features`].
    pub fn features<I, S>(mut self, features: I) -> Self
    where
        I: IntoIterator<Item = S>,
        S: Into<String>,
    {
        self.features = features.into_iter().map(Into::into).collect();
        self
    }

    /// Controls whether the source dependency's default features are enabled.
    /// Defaults to `true`.
    pub fn default_features(mut self, enabled: bool) -> Self {
        self.no_default_features = !enabled;
        self
    }

    /// Adds a source Rust type and its generated opaque struct type.
    ///
    /// `rust_type` must be resolvable from the probe crate. `opaque_type` must
    /// contain one unqualified Rust identifier.
    pub fn add(mut self, rust_type: syn::Type, opaque_type: syn::Type) -> Self {
        use quote::ToTokens;
        self.types.push(OpaqueType::new(
            rust_type.to_token_stream().to_string(),
            opaque_type.to_token_stream().to_string(),
        ));
        self
    }

    /// Override the `Cargo.lock` copied into the probe crate (default: the
    /// destination workspace's lock, located via
    /// [`get-cargo-lock`](https://crates.io/crates/get-cargo-lock)).
    pub fn cargo_lock(mut self, path: impl Into<PathBuf>) -> Self {
        self.cargo_lock = Some(path.into());
        self
    }

    /// Override the probe build directory (default: `$OUT_DIR/opaque_probe`).
    pub fn build_dir(mut self, path: impl Into<PathBuf>) -> Self {
        self.build_dir = Some(path.into());
        self
    }

    /// Probes every requested type and writes the generated Rust source to
    /// `destination`.
    ///
    /// The destination is written only after every requested type's size and
    /// alignment has been read successfully. Failure to build or inspect any
    /// requested type returns an error and leaves an existing destination file
    /// unchanged.
    pub fn generate(&self, destination: impl AsRef<Path>) -> Result<()> {
        if self.types.is_empty() {
            bail!("no opaque types were requested");
        }
        let build_dir = self
            .build_dir
            .clone()
            .ok_or_else(|| anyhow!("build_dir not set and OUT_DIR is unavailable"))?;
        let target = std::env::var("TARGET").unwrap_or_default();
        write_probe_crate(self, &build_dir)?;
        let rlib = build_probe(self, &build_dir, &target)?;
        let data = std::fs::read(&rlib).with_context(|| format!("reading {}", rlib.display()))?;

        validate_types(&self.types)?;
        let mut out = String::from("// @generated by opaque-types — do not edit.\n\n");
        for t in &self.types {
            let size = read_symbol_usize(&data, &sym_name("SIZE", &t.opaque_name))
                .with_context(|| format!("probing size of `{}`", t.rust_path))?;
            let align = read_symbol_usize(&data, &sym_name("ALIGN", &t.opaque_name))
                .with_context(|| format!("probing align of `{}`", t.rust_path))?;
            out.push_str(&render_opaque(&t.opaque_name, size, align));
        }
        let destination = destination.as_ref();
        if let Some(parent) = destination.parent() {
            std::fs::create_dir_all(parent)
                .with_context(|| format!("creating destination directory {}", parent.display()))?;
        }
        std::fs::write(destination, out)
            .with_context(|| format!("writing {}", destination.display()))?;
        Ok(())
    }
}

const PROBE_CRATE: &str = "opaque_types_probe";

/// Symbol name for a probed quantity (`"SIZE"` / `"ALIGN"`) of an opaque type.
fn sym_name(kind: &str, opaque_name: &str) -> String {
    format!("OPAQUE_TYPES_{kind}_{opaque_name}")
}

fn validate_types(types: &[OpaqueType]) -> Result<()> {
    for mapping in types {
        syn::parse_str::<syn::Type>(&mapping.rust_path)
            .with_context(|| format!("invalid Rust type expression `{}`", mapping.rust_path))?;
        syn::parse_str::<syn::Ident>(&mapping.opaque_name).with_context(|| {
            format!(
                "invalid opaque struct name `{}`: expected one Rust identifier",
                mapping.opaque_name
            )
        })?;
    }
    Ok(())
}

/// Render the probe crate's `lib.rs`: one `#[no_mangle] static usize` per quantity.
pub fn render_probe_lib(types: &[OpaqueType]) -> String {
    let mut s = String::from(
        "// @generated probe crate for opaque-types — do not edit.\n\
         #![allow(non_upper_case_globals, dead_code)]\n",
    );
    for t in types {
        let size_sym = sym_name("SIZE", &t.opaque_name);
        let align_sym = sym_name("ALIGN", &t.opaque_name);
        let path = &t.rust_path;
        s.push_str(&format!(
            "#[no_mangle]\n#[used]\npub static {size_sym}: usize = ::core::mem::size_of::<{path}>();\n\
             #[no_mangle]\n#[used]\npub static {align_sym}: usize = ::core::mem::align_of::<{path}>();\n",
        ));
    }
    s
}

/// Renders one `#[repr(C, align)]` opaque storage struct.
///
/// The result has the supplied size and alignment but defines no conversion
/// behavior or representation invariant.
pub fn render_opaque(opaque_name: &str, size: usize, align: usize) -> String {
    format!(
        "#[repr(C, align({align}))]\n#[allow(non_camel_case_types)]\n\
         pub struct {opaque_name} {{\n    pub _0: [u8; {size}],\n}}\n\n"
    )
}

/// The consuming project's `Cargo.lock`, so the probe uses the same resolution.
///
/// The path comes from [`get_cargo_lock::get_cargo_lock`]. This dependency is
/// patched to the proxy installed in the destination workspace, allowing this
/// externally developed crate to obtain that workspace's lockfile. The consumer
/// must have run `cargo get-cargo-lock install` (otherwise `get_cargo_lock`
/// panics with guidance — there is intentionally no silent fallback).
fn default_cargo_lock() -> PathBuf {
    get_cargo_lock::get_cargo_lock()
}

/// Read the `[package].name` of the crate whose manifest dir is `manifest_dir`.
/// It is the `[dependencies]` key the probe must use for a path dependency.
fn read_package_name(manifest_dir: &Path) -> Result<String> {
    let manifest_path = manifest_dir.join("Cargo.toml");
    let text = std::fs::read_to_string(&manifest_path)
        .with_context(|| format!("reading {}", manifest_path.display()))?;
    let table: toml::Table = text
        .parse()
        .with_context(|| format!("parsing {}", manifest_path.display()))?;
    table
        .get("package")
        .and_then(|p| p.get("name"))
        .and_then(|n| n.as_str())
        .map(str::to_string)
        .ok_or_else(|| {
            anyhow!(
                "no `[package].name` (string) in {}",
                manifest_path.display()
            )
        })
}

/// Write the probe crate (`Cargo.toml` + `src/lib.rs`, and `Cargo.lock` if given).
fn write_probe_crate(b: &OpaqueTypes, build_dir: &Path) -> Result<()> {
    let src = build_dir.join("src");
    std::fs::create_dir_all(&src)
        .with_context(|| format!("creating probe src dir {}", src.display()))?;
    let source_manifest_dir = b.source_manifest_dir.canonicalize().with_context(|| {
        format!(
            "resolving source manifest directory {}",
            b.source_manifest_dir.display()
        )
    })?;
    let package = read_package_name(&source_manifest_dir)?;
    let source_path = source_manifest_dir.to_str().ok_or_else(|| {
        anyhow!(
            "source manifest directory is not valid UTF-8: {}",
            source_manifest_dir.display()
        )
    })?;

    let features_toml = if b.features.is_empty() {
        String::new()
    } else {
        let list = b
            .features
            .iter()
            .map(|feature| toml::Value::String(feature.clone()).to_string())
            .collect::<Vec<_>>()
            .join(", ");
        format!(", features = [{list}]")
    };
    let manifest = format!(
        "[package]\nname = \"{PROBE_CRATE}\"\nversion = \"0.0.0\"\nedition = \"2021\"\n\
         publish = false\n\n[lib]\ncrate-type = [\"lib\"]\n\n[dependencies]\n\
         {pkg} = {{ path = {path}, default-features = {dflt}{features} }}\n\n\
         [workspace]\n",
        pkg = toml::Value::String(package),
        path = toml::Value::String(source_path.to_owned()),
        dflt = !b.no_default_features,
        features = features_toml,
    );
    std::fs::write(build_dir.join("Cargo.toml"), manifest)?;
    std::fs::write(src.join("lib.rs"), render_probe_lib(&b.types))?;
    let cargo_lock = b.cargo_lock.clone().unwrap_or_else(default_cargo_lock);
    std::fs::copy(&cargo_lock, build_dir.join("Cargo.lock"))
        .with_context(|| format!("copying lockfile {} into probe", cargo_lock.display()))?;
    Ok(())
}

/// Build the probe crate for `$TARGET` and return the path to its rlib.
fn build_probe(_b: &OpaqueTypes, build_dir: &Path, target: &str) -> Result<PathBuf> {
    let mut cmd = std::process::Command::new(std::env::var("CARGO").unwrap_or("cargo".into()));
    cmd.current_dir(build_dir)
        .arg("build")
        .arg("--offline")
        .arg("--message-format=json-render-diagnostics")
        .arg("--manifest-path")
        .arg(build_dir.join("Cargo.toml"));
    if !target.is_empty() {
        cmd.arg("--target").arg(target);
    }
    // Isolate the probe's target dir from the consumer's (avoid lock contention).
    cmd.arg("--target-dir").arg(build_dir.join("target"));
    let out = cmd.output().context("spawning cargo for the probe crate")?;
    if !out.status.success() {
        bail!(
            "probe build failed:\nstdout:\n{}\nstderr:\n{}",
            String::from_utf8_lossy(&out.stdout),
            String::from_utf8_lossy(&out.stderr),
        );
    }
    // Parse cargo JSON for the probe crate's rlib artifact.
    let stdout = String::from_utf8_lossy(&out.stdout);
    let mut rlib: Option<PathBuf> = None;
    for line in stdout.lines() {
        // Minimal JSON scan (avoid a serde dep): look for the probe's artifact line.
        if line.contains("\"compiler-artifact\"") && line.contains(PROBE_CRATE) {
            if let Some(p) = extract_first_rlib(line) {
                rlib = Some(PathBuf::from(p));
            }
        }
    }
    rlib.ok_or_else(|| anyhow!("probe rlib artifact not found in cargo output"))
}

/// Pull the first `.rlib` path out of a cargo `compiler-artifact` JSON line.
fn extract_first_rlib(line: &str) -> Option<String> {
    // `"filenames":["...rlib", ...]` — find the first quoted token ending in .rlib.
    // Cargo emits JSON, so on Windows the path's backslashes arrive escaped
    // (`C:\\…\\libprobe.rlib`); unescape the standard JSON sequences before use.
    let idx = line.find("\"filenames\"")?;
    let rest = &line[idx..];
    for tok in rest.split('"') {
        if tok.ends_with(".rlib") {
            return Some(json_unescape(tok));
        }
    }
    None
}

/// Minimal JSON string unescaping for the path tokens cargo emits (`\\`, `\"`,
/// `\/`). Sufficient for filesystem paths; not a general JSON unescaper.
fn json_unescape(s: &str) -> String {
    let mut out = String::with_capacity(s.len());
    let mut chars = s.chars();
    while let Some(c) = chars.next() {
        if c == '\\' {
            match chars.next() {
                Some('\\') => out.push('\\'),
                Some('"') => out.push('"'),
                Some('/') => out.push('/'),
                Some(other) => {
                    out.push('\\');
                    out.push(other);
                }
                None => out.push('\\'),
            }
        } else {
            out.push(c);
        }
    }
    out
}

/// Read the `usize` value of a `#[no_mangle] static` named `sym` from a compiled
/// rlib/object (an `ar` archive of object files). Tries the bare name and the
/// Mach-O `_`-prefixed variant.
pub fn read_symbol_usize(artifact: &[u8], sym: &str) -> Result<usize> {
    use object::{Object, ObjectSection, ObjectSymbol};

    let with_underscore = format!("_{sym}");
    let matches = |name: &str| name == sym || name == with_underscore;

    let read_from = |obj: &object::File| -> Option<usize> {
        let ptr_bytes = if obj.is_64() { 8 } else { 4 };
        let s = obj
            .symbols()
            .find(|s| s.name().map(matches).unwrap_or(false))?;
        let sec = obj.section_by_index(s.section_index()?).ok()?;
        let data = sec.data().ok()?;
        let off = s.address().checked_sub(sec.address())? as usize;
        let bytes = data.get(off..off + ptr_bytes)?;
        let value = match (ptr_bytes, obj.is_little_endian()) {
            (4, true) => u32::from_le_bytes(bytes.try_into().ok()?) as u64,
            (4, false) => u32::from_be_bytes(bytes.try_into().ok()?) as u64,
            (8, true) => u64::from_le_bytes(bytes.try_into().ok()?),
            (8, false) => u64::from_be_bytes(bytes.try_into().ok()?),
            _ => return None,
        };
        usize::try_from(value).ok()
    };

    // rlib / .a is an ar archive of object members; plain .o parses directly.
    if let Ok(archive) = object::read::archive::ArchiveFile::parse(artifact) {
        for member in archive.members() {
            let member = member.map_err(|e| anyhow!("archive member: {e}"))?;
            let data = member
                .data(artifact)
                .map_err(|e| anyhow!("archive member data: {e}"))?;
            if let Ok(obj) = object::File::parse(data) {
                if let Some(v) = read_from(&obj) {
                    return Ok(v);
                }
            }
        }
    } else if let Ok(obj) = object::File::parse(artifact) {
        if let Some(v) = read_from(&obj) {
            return Ok(v);
        }
    }
    bail!("symbol `{sym}` not found in probe artifact")
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn probe_lib_emits_size_and_align_statics() {
        let types = vec![OpaqueType::new("model::Message", "message_t")];
        let s = render_probe_lib(&types);
        assert!(s.contains(
            "pub static OPAQUE_TYPES_SIZE_message_t: usize = ::core::mem::size_of::<model::Message>();"
        ));
        assert!(s.contains(
            "pub static OPAQUE_TYPES_ALIGN_message_t: usize = ::core::mem::align_of::<model::Message>();"
        ));
        assert!(s.contains("#[no_mangle]") && s.contains("#[used]"));
    }

    #[test]
    fn features_are_explicit_and_independent_of_defaults() {
        let b = OpaqueTypes::new("source")
            .features(["unstable", "shared-memory"])
            .add(
                syn::parse_quote!(model::Message),
                syn::parse_quote!(message_t),
            );
        assert_eq!(b.features, vec!["unstable", "shared-memory"]);
        assert!(!b.no_default_features);
        assert_eq!(b.types.len(), 1);
        assert_eq!(b.types[0].opaque_name, "message_t");
    }

    #[test]
    fn invalid_mapping_is_rejected() {
        let mappings = [OpaqueType::new("not a type!", "not::an::identifier")];
        let error = validate_types(&mappings).unwrap_err().to_string();
        assert!(error.contains("invalid Rust type expression"));
    }

    #[test]
    fn opaque_struct_renders_repr_c_align() {
        let s = render_opaque("z_zbytes_t", 32, 8);
        assert!(s.contains("#[repr(C, align(8))]"));
        assert!(s.contains("pub struct z_zbytes_t"));
        assert!(s.contains("pub _0: [u8; 32]"));
    }

    #[test]
    fn rlib_artifact_path_parsed_from_cargo_json() {
        let line = r#"{"reason":"compiler-artifact","package_id":"opaque_types_probe 0.0.0","filenames":["/tmp/t/target/debug/deps/libopaque_types_probe-abc.rlib"],"executable":null}"#;
        assert_eq!(
            extract_first_rlib(line).as_deref(),
            Some("/tmp/t/target/debug/deps/libopaque_types_probe-abc.rlib")
        );
    }

    #[test]
    fn rlib_artifact_path_windows_backslashes_unescaped() {
        // Cargo JSON escapes Windows backslashes as `\\`.
        let line = r#"{"reason":"compiler-artifact","filenames":["C:\\proj\\target\\debug\\deps\\libopaque_types_probe-abc.rlib"]}"#;
        assert_eq!(
            extract_first_rlib(line).as_deref(),
            Some(r"C:\proj\target\debug\deps\libopaque_types_probe-abc.rlib")
        );
    }

    #[test]
    fn generates_layout_from_a_temporary_source_package() -> Result<()> {
        let temporary = tempfile::tempdir()?;
        let source = temporary.path().join("model");
        std::fs::create_dir_all(source.join("src"))?;
        std::fs::write(
            source.join("Cargo.toml"),
            "[package]\nname = \"layout-model\"\nversion = \"0.0.0\"\nedition = \"2021\"\n",
        )?;
        std::fs::write(
            source.join("src/lib.rs"),
            "#[repr(C, align(16))]\npub struct Value(pub [u8; 3]);\n",
        )?;
        let lockfile = source.join("Cargo.lock");
        std::fs::write(&lockfile, "version = 4\n")?;

        let destination = temporary.path().join("generated/opaque_types.rs");
        OpaqueTypes::new(&source)
            .cargo_lock(lockfile)
            .build_dir(temporary.path().join("probe"))
            .add(
                syn::parse_quote!(layout_model::Value),
                syn::parse_quote!(opaque_value_t),
            )
            .generate(&destination)?;

        let generated = std::fs::read_to_string(destination)?;
        assert!(generated.contains("#[repr(C, align(16))]"));
        assert!(generated.contains("pub struct opaque_value_t"));
        assert!(generated.contains("pub _0: [u8; 16]"));
        Ok(())
    }

    #[test]
    fn failed_probe_does_not_replace_destination() -> Result<()> {
        let temporary = tempfile::tempdir()?;
        let source = temporary.path().join("model");
        std::fs::create_dir_all(source.join("src"))?;
        std::fs::write(
            source.join("Cargo.toml"),
            "[package]\nname = \"layout-model\"\nversion = \"0.0.0\"\nedition = \"2021\"\n",
        )?;
        std::fs::write(source.join("src/lib.rs"), "pub struct Value;\n")?;
        let lockfile = source.join("Cargo.lock");
        std::fs::write(&lockfile, "version = 4\n")?;
        let destination = temporary.path().join("opaque_types.rs");
        std::fs::write(&destination, "existing output\n")?;

        let result = OpaqueTypes::new(&source)
            .cargo_lock(lockfile)
            .build_dir(temporary.path().join("probe"))
            .add(
                syn::parse_quote!(layout_model::Value),
                syn::parse_quote!(opaque_value_t),
            )
            .add(
                syn::parse_quote!(layout_model::Missing),
                syn::parse_quote!(missing_t),
            )
            .generate(&destination);

        assert!(result.is_err());
        assert_eq!(std::fs::read_to_string(destination)?, "existing output\n");
        Ok(())
    }
}