generate corpus to fuzz on with meson

3 files changed, 83 insertions, 2 deletions

diff --git a/fuzz/fuzz.rs b/fuzz/fuzz.rs
index 69804e8..72f751b 100644
--- a/fuzz/fuzz.rs
+++ b/fuzz/fuzz.rs
@@ -19,7 +19,7 @@ pub unsafe extern "C" fn LLVMFuzzerTestOneInput(input: *const u8, len: usize) ->
     let mut proc = Command::new("atom.py")
         .stdin(Stdio::piped())
         .spawn()
-        .unwrap();
+        .expect("failed to start atom.py");
 
     proc.stdin
         .as_mut()
diff --git a/fuzz/gencorpus.rs b/fuzz/gencorpus.rs
new file mode 100644
index 0000000..67ba1e3
--- /dev/null
+++ b/fuzz/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/meson.build b/fuzz/meson.build
index c4b5e8d..d6809f6 100644
--- a/fuzz/meson.build
+++ b/fuzz/meson.build
@@ -1,5 +1,20 @@
 cbindgen = find_program('cbindgen')
 
+gencorpus = executable(
+    'gencorpus',
+    'gencorpus.rs',
+    dependencies: [mon],
+    link_with: [gentoo_utils],
+)
+
+corpus_directory = meson.current_build_dir() / 'corpus'
+
+corpus = custom_target(
+    'corpus',
+    output: 'corpus',
+    command: [gencorpus, corpus_directory],
+)
+
 fuzz_h = custom_target(
     'fuzz_h',
     input: 'fuzz.rs',
@@ -20,8 +35,10 @@ fuzz_rs = static_library(
     link_with: [gentoo_utils],
 )
 
-fuzz_cpp = executable(
+fuzz = executable(
     'fuzz',
     link_args: ['-fsanitize=fuzzer'],
     link_with: [fuzz_rs],
 )
+
+test('fuzz', fuzz, args: [corpus_directory], depends: [corpus], timeout: 0)