3 files changed, 170 insertions, 0 deletions
diff --git a/fuzz/atom/parser/fuzz.rs b/fuzz/atom/parser/fuzz.rs
new file mode 100644
index 0000000..610e08b
--- /dev/null
+++ b/fuzz/atom/parser/fuzz.rs
@@ -0,0 +1,100 @@
+use core::slice;
+use gentoo_utils::{Parseable, atom::Atom};
+use mon::{Parser, ParserFinishedError, input::InputIter};
+use std::{
+    io::{BufRead, BufReader, Write},
+    process::{ChildStdin, ChildStdout, Command, Stdio},
+    sync::{LazyLock, Mutex},
+};
+
+struct PyProcess {
+    stdin: Mutex<ChildStdin>,
+    stdout: Mutex<BufReader<ChildStdout>>,
+    buffer: Mutex<String>,
+}
+
+#[allow(clippy::missing_safety_doc, clippy::needless_return)]
+#[unsafe(no_mangle)]
+pub unsafe extern "C" fn LLVMFuzzerTestOneInput(input: *const u8, len: usize) -> i32 {
+    static PY_PROCESS: LazyLock<PyProcess> = LazyLock::new(|| {
+        #[allow(clippy::zombie_processes)]
+        let mut proc = Command::new("atom.py")
+            .stdin(Stdio::piped())
+            .stdout(Stdio::piped())
+            .stderr(Stdio::inherit())
+            .spawn()
+            .expect("failed to spawn atom.py");
+
+        let stdin = Mutex::new(proc.stdin.take().unwrap());
+        let stdout = Mutex::new(BufReader::new(proc.stdout.take().unwrap()));
+
+        PyProcess {
+            stdin,
+            stdout,
+            buffer: Mutex::new(String::new()),
+        }
+    });
+
+    let slice = unsafe { slice::from_raw_parts(input, len) };
+
+    if slice.iter().any(|b| !b.is_ascii_graphic()) {
+        return -1;
+    }
+
+    let str = match str::from_utf8(slice) {
+        Ok(str) => str,
+        Err(_) => return -1,
+    };
+
+    let atom = str.trim();
+
+    let mut stdin = PY_PROCESS.stdin.lock().expect("failed to get stdin lock");
+
+    writeln!(&mut stdin, "{atom}").expect("failed to write to python stdin");
+
+    let mut stdout = PY_PROCESS.stdout.lock().expect("failed to get stdout lock");
+
+    let mut buffer = PY_PROCESS.buffer.lock().expect("failed to get buffer lock");
+
+    buffer.clear();
+
+    stdout
+        .read_line(&mut buffer)
+        .expect("failed to readline from python");
+
+    let portage_result = match buffer.as_str().trim() {
+        "0" => true,
+        "1" => false,
+        result => panic!("got unexpected result from python: {result}"),
+    };
+
+    let gentoo_utils_result = Atom::parser().parse_finished(InputIter::new(atom));
+
+    match (portage_result, gentoo_utils_result) {
+        (true, Ok(_)) => {
+            eprintln!("agreement that {atom} is valid");
+        }
+        (false, Err(_)) => {
+            eprintln!("agreement that {atom} is invalid");
+        }
+        (true, Err(_)) => {
+            panic!("rejected valid atom: {atom}");
+        }
+        (false, Ok(atom))
+            if atom.usedeps().iter().any(|usedep| {
+                atom.usedeps()
+                    .iter()
+                    .filter(|u| usedep.flag() == u.flag())
+                    .count()
+                    > 1
+            }) =>
+        {
+            eprintln!("disagreement due to duplicates in usedeps");
+        }
+        (false, Ok(_)) => {
+            panic!("accpeted invalid atom: {atom}")
+        }
+    }
+
+    return 0;
+}
diff --git a/fuzz/atom/parser/gencorpus.rs b/fuzz/atom/parser/gencorpus.rs
new file mode 100644
index 0000000..67ba1e3
--- /dev/null
+++ b/fuzz/atom/parser/gencorpus.rs
@@ -0,0 +1,64 @@
+use std::{
+    env,
+    error::Error,
+    fs::{self, OpenOptions},
+    io::Write,
+    path::PathBuf,
+};
+
+use gentoo_utils::{
+    atom::Atom,
+    ebuild::{Depend, repo::Repo},
+};
+
+fn main() -> Result<(), Box<dyn Error>> {
+    let corpus_dir = PathBuf::from(
+        env::args()
+            .nth(1)
+            .expect("expected corpus directory as first argument"),
+    );
+
+    fs::create_dir_all(&corpus_dir)?;
+
+    let repo = Repo::new("/var/db/repos/gentoo");
+    let mut atoms = Vec::new();
+
+    for category in repo.categories()? {
+        for ebuild in category?.ebuilds()? {
+            let depend = ebuild?.depend().to_vec();
+
+            for expr in depend {
+                walk_expr(&mut atoms, &expr);
+            }
+        }
+    }
+
+    for (i, atom) in atoms.iter().enumerate() {
+        let path = corpus_dir.as_path().join(i.to_string());
+        let mut file = OpenOptions::new()
+            .write(true)
+            .truncate(true)
+            .create(true)
+            .open(path)?;
+
+        write!(file, "{atom}")?;
+    }
+
+    Ok(())
+}
+
+fn walk_expr(atoms: &mut Vec<Atom>, depend: &Depend<Atom>) {
+    match depend {
+        Depend::Element(atom) => {
+            atoms.push(atom.clone());
+        }
+        Depend::AllOf(exprs)
+        | Depend::OneOf(exprs)
+        | Depend::AnyOf(exprs)
+        | Depend::ConditionalGroup(_, exprs) => {
+            for expr in exprs {
+                walk_expr(atoms, expr);
+            }
+        }
+    }
+}
diff --git a/fuzz/atom/parser/meson.build b/fuzz/atom/parser/meson.build
new file mode 100644
index 0000000..efc84c6
--- /dev/null
+++ b/fuzz/atom/parser/meson.build
@@ -0,0 +1,6 @@
+fuzzers += {
+    'atom_parser': [
+        meson.current_source_dir() / 'gencorpus.rs',
+        meson.current_source_dir() / 'fuzz.rs',
+    ],
+}