Keep isolated document extraction working on macOS

crates/searchi-core/src/extraction_process.rs

@@ -48,6 +48,33 @@ enum Event {
     Stderr,
 }
 
+#[cfg(target_os = "macos")]
+fn check_resident_memory(child: &mut Child, maximum: u64) -> Result<(), ProcessError> {
+    let mut info = std::mem::MaybeUninit::<libc::proc_taskinfo>::uninit();
+    let size = std::mem::size_of::<libc::proc_taskinfo>();
+    // The child stays owned and unreaped until I/O completes. A failed query is
+    // accepted only when it has already exited; otherwise fail closed.
+    let read = unsafe {
+        libc::proc_pidinfo(
+            child.id() as i32,
+            libc::PROC_PIDTASKINFO,
+            0,
+            info.as_mut_ptr().cast(),
+            size as i32,
+        )
+    };
+    if read != size as i32 {
+        if child.try_wait().map_err(failed)?.is_some() {
+            return Ok(());
+        }
+        return Err(failed("Cannot inspect extraction memory usage"));
+    }
+    if unsafe { info.assume_init() }.pti_resident_size > maximum {
+        return Err(failed("Extraction worker resident memory exceeds budget"));
+    }
+    Ok(())
+}
+
 /// Includes pipe writes, response EOF and process exit in the same deadline.
 /// No parser fallback occurs when limits, transport or worker execution fail.
 pub fn run(
@@ -161,6 +188,8 @@ pub fn run(
         if started.elapsed() >= config.timeout {
             return Err(ProcessError::TimedOut);
         }
+        #[cfg(target_os = "macos")]
+        check_resident_memory(&mut owned.child, 512 * 1024 * 1024)?;
         match rx.recv_timeout(
             Duration::from_millis(5).min(config.timeout.saturating_sub(started.elapsed())),
         ) {
@@ -288,6 +317,20 @@ mod tests {
             .unwrap_or_else(|poisoned| poisoned.into_inner())
     }
 
+    #[cfg(target_os = "macos")]
+    #[test]
+    fn mac_memory_budget_rejects_a_live_owned_child() {
+        let mut child = Command::new("/bin/sleep").arg("30").spawn().unwrap();
+        // Wait for exec before measuring, then impose a deliberately tiny budget.
+        thread::sleep(Duration::from_millis(50));
+        let result = check_resident_memory(&mut child, 1);
+        let _ = child.kill();
+        child.wait().unwrap();
+        assert!(
+            matches!(result, Err(ProcessError::Failed(message)) if message.contains("exceeds budget"))
+        );
+    }
+
     fn fixtures() -> tempfile::TempDir {
         let directory = tempfile::tempdir().unwrap();
         let source = directory.path().join("fake.rs");

crates/searchi-extract-worker/src/resources.rs

@@ -1,5 +1,6 @@
 use anyhow::{ensure, Result};
 
+#[cfg(not(target_os = "macos"))]
 const MEMORY_BYTES: usize = 512 * 1024 * 1024;
 
 #[cfg(windows)]
@@ -37,7 +38,10 @@ pub fn apply() -> Result<()> {
 
 #[cfg(unix)]
 pub fn apply() -> Result<()> {
+    // Darwin rejects finite AS/DATA limits. The parent enforces a sampled RSS
+    // budget there; retain the CPU limit and bounded path-free transport here.
     for (resource, maximum) in [
+        #[cfg(not(target_os = "macos"))]
         (libc::RLIMIT_AS, MEMORY_BYTES as libc::rlim_t),
         (libc::RLIMIT_CPU, 30 as libc::rlim_t),
     ] {