Files
searchi/crates/searchi-core/tests/document_publication.rs
T

108 lines
4.4 KiB
Rust

//! Deterministic parent/index publication tests using an owned protocol fixture.
//! Genuine parser execution is covered separately in documents.rs.
use anyhow::Result;
use searchi_core::{
extraction_protocol::{DocumentFormat, RequestHeader},
IndexOptions, SearchIndex,
};
use std::{
fs,
sync::atomic::{AtomicBool, Ordering},
time::{Duration, Instant},
};
#[test]
fn mid_worker_same_stat_changes_and_cancellation_roll_back_index() -> Result<()> {
for cancel in [false, true] {
let directory = tempfile::tempdir()?;
let root = directory.path().join("files");
fs::create_dir(&root)?;
let source = root.join("contract.pdf");
fs::write(root.join("notes.txt"), "preserved evidence")?;
let mut index = SearchIndex::open(":memory:")?;
index.index_folder(&root)?;
fs::write(&source, b"original")?;
let initial = fs::metadata(&source)?;
let digest = RequestHeader::from_bytes(DocumentFormat::Pdf, b"original")?.input_sha256;
let marker = directory.path().join("admitted");
let release = directory.path().join("release");
let fixture_source = directory.path().join("worker.rs");
let executable = directory
.path()
.join(format!("fixture{}", std::env::consts::EXE_SUFFIX));
let response = serde_json::json!({
"version":1,"input_sha256":digest,
"outcome":{"status":"extracted","passages":[{"text":"stale document text","source":{"kind":"page","number":1}}]}
}).to_string();
fs::write(
&fixture_source,
format!(
r#"
use std::io::{{Read,Write}};
fn main() {{
let mut bytes=Vec::new(); std::io::stdin().read_to_end(&mut bytes).unwrap();
std::fs::write({marker:?}, b"admitted").unwrap();
while !std::path::Path::new({release:?}).exists() {{ std::thread::sleep(std::time::Duration::from_millis(5)); }}
let response={response:?}.as_bytes(); let mut out=std::io::stdout();
out.write_all(&(response.len() as u32).to_le_bytes()).unwrap(); out.write_all(response).unwrap();
}}
"#
),
)?;
let status = std::process::Command::new("rustc")
.arg(&fixture_source)
.arg("-o")
.arg(&executable)
.status()?;
anyhow::ensure!(status.success(), "Compile owned protocol fixture");
index.set_document_worker(executable);
let cancelled = AtomicBool::new(false);
std::thread::scope(|scope| {
let changer = scope.spawn(|| -> Result<()> {
let deadline = Instant::now() + Duration::from_secs(10);
while !marker.exists() && Instant::now() < deadline {
std::thread::sleep(Duration::from_millis(5));
}
if !marker.exists() {
cancelled.store(true, Ordering::Release);
anyhow::bail!("Worker did not admit fixture input");
}
if cancel {
cancelled.store(true, Ordering::Release);
} else {
fs::write(&source, b"modified")?;
fs::File::options()
.write(true)
.open(&source)?
.set_times(fs::FileTimes::new().set_modified(initial.modified()?))?;
assert_eq!(fs::metadata(&source)?.len(), initial.len());
assert_eq!(fs::metadata(&source)?.modified()?, initial.modified()?);
fs::write(&release, b"release")?;
}
Ok(())
});
let result =
index.index_folder_cancellable(&root, &IndexOptions::default(), &cancelled);
assert!(
result.is_err(),
"Mutation or cancellation must roll back publication"
);
let message = format!("{:#}", result.unwrap_err());
assert!(
message.contains(if cancel {
"cancelled"
} else {
"changed during extraction"
}),
"{message}"
);
changer.join().unwrap().unwrap();
});
assert_eq!(index.stats()?.files, 1);
assert_eq!(index.search("evidence", 10)?.len(), 1);
assert!(index.search("stale", 10)?.is_empty());
assert!(!index.contains_path(&source.to_string_lossy())?);
}
Ok(())
}