From 7e3824c0bc1cfe9bdfa4c9471725e77fdc9718e2 Mon Sep 17 00:00:00 2001 From: jyviko <2501648+jyviko@users.noreply.github.com> Date: Tue, 3 Nov 2026 06:24:51 -0500 Subject: [PATCH] Kernel rig export: operand records in a gzip file beside each entry An entry keeps what each operand body is and names .bodies.json.gz, which holds the records (gzip with no timestamp, so the same records give the same bytes); the entry itself is pretty JSON. The corpus of 1202 cases is 33 MB. Co-Authored-By: Claude Opus 5.5 Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ --- Cargo.lock | 1 + xtask/Cargo.toml | 1 + xtask/src/kernel_corpus.rs | 30 ++++++++++++++++++++++++++++-- 3 files changed, 30 insertions(+), 2 deletions(-) diff --git a/Cargo.lock b/Cargo.lock index ccec6b015..efcd6f5da 100644 --- a/Cargo.lock +++ b/Cargo.lock @@ -5754,6 +5754,7 @@ checksum = "e450f9b2ed1dff33c94c12589a87338689467b9c4f5d8a5710bd09a847d2c8a7" name = "xtask" version = "0.40.0-rc.1" dependencies = [ + "flate2", "proc-macro2", "quick-xml", "serde", diff --git a/xtask/Cargo.toml b/xtask/Cargo.toml index 452087caa..14455fd21 100644 --- a/xtask/Cargo.toml +++ b/xtask/Cargo.toml @@ -6,6 +6,7 @@ license.workspace = true publish = false [dependencies] +flate2 = "1.1" proc-macro2 = { workspace = true, features = ["span-locations"] } quick-xml = { workspace = true } serde = { workspace = true } diff --git a/xtask/src/kernel_corpus.rs b/xtask/src/kernel_corpus.rs index fb366917a..d3364bfb6 100644 --- a/xtask/src/kernel_corpus.rs +++ b/xtask/src/kernel_corpus.rs @@ -4,7 +4,9 @@ //! the fix set, its expected values and the rig's result). This writes what that leaves out: //! an entry for each case whose process left none (a hang or a crash, so no trace), each ladder //! case's golden, and `DIR/manifest.json`, the counts per area and every case that is not -//! expressible as kernel input, with its reasons. +//! expressible as kernel input, with its reasons. An entry's operand records go to +//! `.bodies.json.gz` beside it (gzip, no timestamp, so a re-export of the same records is +//! the same bytes); the entry keeps what each body is and names that file. use std::collections::BTreeMap; use std::path::Path; @@ -43,6 +45,29 @@ fn write_json(path: &Path, v: &Value, pretty: bool) -> Result<(), String> { std::fs::write(path, text).map_err(|e| format!("{}: {e}", path.display())) } +/// Moves the records out of `entry`'s `bodies` into the gzip file `to`, `{"b": record}`, and +/// names it in `records`. An entry without records (written before) is left as it is. +fn split_records(entry: &mut Value, to: &Path) -> Result<(), String> { + use std::io::Write as _; + let Some(bodies) = entry["bodies"].as_object_mut() else { return Ok(()) }; + let mut records = serde_json::Map::new(); + for (name, b) in bodies.iter_mut() { + if let Some(r) = b.as_object_mut().and_then(|o| o.remove("record")) { + records.insert(name.clone(), r); + } + } + if records.is_empty() { + return Ok(()); + } + let text = serde_json::to_vec(&Value::Object(records)).map_err(|e| e.to_string())?; + let mut gz = flate2::write::GzEncoder::new(Vec::new(), flate2::Compression::best()); + gz.write_all(&text).map_err(|e| e.to_string())?; + let bytes = gz.finish().map_err(|e| e.to_string())?; + std::fs::write(to, bytes).map_err(|e| format!("{}: {e}", to.display()))?; + entry["records"] = json!(to.file_name().map(|n| n.to_string_lossy().to_string())); + Ok(()) +} + /// What `git args` prints in this checkout. fn git(args: &[&str]) -> String { std::process::Command::new("git") @@ -95,7 +120,8 @@ pub(crate) fn manifest(dir: &Path, results: &[Value], golden: Option<&Value>) -> if let Some(e) = r.get("export_error") { entry["export_error"] = e.clone(); } - write_json(&file, &entry, false)?; + split_records(&mut entry, &dir.join(area).join(format!("{id}.bodies.json.gz")))?; + write_json(&file, &entry, true)?; } // Every entry in the corpus, this run's or an earlier one's.