[pitboss/grind] deferred session-0009 (20260520T233019Z-6958)

This commit is contained in:
pitboss 2026-05-21 03:39:36 -05:00
parent a6f34554db
commit 38cc0ce05f
60 changed files with 509 additions and 541 deletions

View file

@ -1,45 +1,44 @@
/// Dynamic verification benchmarks (§8.4). //! Dynamic verification benchmarks (§8.4).
/// //!
/// Tracks the per-scan cost anchors: //! Tracks the per-scan cost anchors:
/// //!
/// 1. `harness_build_cold` — fresh workdir, spec → BuiltHarness (source gen + disk write). //! 1. `harness_build_cold` — fresh workdir, spec → BuiltHarness (source gen + disk write).
/// 2. `harness_build_warm` — same spec, workdir already staged (file write skipped). //! 2. `harness_build_warm` — same spec, workdir already staged (file write skipped).
/// 3. `sandbox_run_payload` — single payload run via process backend against //! 3. `sandbox_run_payload` — single payload run via process backend against
/// sqli_positive.py (subprocess + settrace overhead, no networking). //! sqli_positive.py (subprocess + settrace overhead, no networking).
/// 4. `docker_image_build` — cold image pull/build for the python:3-slim base. //! 4. `docker_image_build` — cold image pull/build for the python:3-slim base.
/// 5. `docker_exec_warm` — `docker exec` into a running container (no cold start). //! 5. `docker_exec_warm` — `docker exec` into a running container (no cold start).
/// 6. `docker_payload_cost` — per-payload sandbox cost via docker backend end-to-end. //! 6. `docker_payload_cost` — per-payload sandbox cost via docker backend end-to-end.
/// 7. `composite_chain_reverify_dispatch` — `reverify_top_chains` on a //! 7. `composite_chain_reverify_dispatch` — `reverify_top_chains` on a
/// synthetic 3-member chain with no member diags. Measures the no-derive //! synthetic 3-member chain with no member diags. Measures the no-derive
/// dispatch path (chain_step_specs miss, early-exit build/run loops, //! dispatch path (chain_step_specs miss, early-exit build/run loops,
/// Inconclusive verdict allocation, severity downgrade). //! Inconclusive verdict allocation, severity downgrade).
/// 8. `composite_chain_reverify_stub_confirmed` — same chain shape, stubbed //! 8. `composite_chain_reverify_stub_confirmed` — same chain shape, stubbed
/// reverifier returning `Confirmed`. Measures the apply-verdict happy path //! reverifier returning `Confirmed`. Measures the apply-verdict happy path
/// (no severity bucket change). //! (no severity bucket change).
/// 9. `composite_chain_reverify_top_n_slice` — 5-chain slice with `top_n=3`. //! 9. `composite_chain_reverify_top_n_slice` — 5-chain slice with `top_n=3`.
/// Measures the slice traversal cost so a regression that walks the full //! Measures the slice traversal cost so a regression that walks the full
/// slice instead of the prefix is visible. //! slice instead of the prefix is visible.
/// 10. `composite_chain_reverify_replay_stable` — same chain shape as //! 10. `composite_chain_reverify_replay_stable` — same chain shape as
/// `stub_confirmed`, but with `VerifyOptions::replay_stable_check=true` //! `stub_confirmed`, but with `VerifyOptions::replay_stable_check=true`
/// and a stub that stamps `replay_stable=Some(true)`. Anchors the //! and a stub that stamps `replay_stable=Some(true)`. Anchors the
/// apply-verdict allocation cost when the telemetry stability field //! apply-verdict allocation cost when the telemetry stability field
/// is populated; a regression that adds per-chain work behind the //! is populated; a regression that adds per-chain work behind the
/// replay opt-in (e.g. an extra run_chain_steps call leaking out of //! replay opt-in (e.g. an extra run_chain_steps call leaking out of
/// the live path into the stub layer) shows up here. //! the live path into the stub layer) shows up here.
/// //!
/// Wall-clock budget anchors for the composite reverify path (per the //! Wall-clock budget anchors for the composite reverify path: the live
/// Phase 26 acceptance literal): the live process backend stays under //! process backend stays under 400ms per 3-member chain, the docker
/// 400ms per 3-member chain, the docker backend under 1500ms. Those //! backend under 1500ms. Those live-run numbers are covered by the
/// live-run numbers are covered by the //! `flask_eval_chain_reverify_populates_dynamic_verdict` integration
/// `flask_eval_chain_reverify_populates_dynamic_verdict` integration //! test in `tests/chain_emission_e2e.rs`; the microbenches here anchor
/// test in `tests/chain_emission_e2e.rs`; the microbenches here anchor //! the dispatch + verdict-application overhead so regressions on the
/// the dispatch + verdict-application overhead so regressions on the //! API-shape half land in the criterion baseline.
/// API-shape half land in the criterion baseline. //!
/// //! Baselines committed to `benches/dynamic_bench_baseline.json`.
/// Baselines committed to `benches/dynamic_bench_baseline.json`. //! Run: `cargo bench --features dynamic -- dynamic`
/// Run: `cargo bench --features dynamic -- dynamic` //!
/// //! Docker benchmarks are no-ops when docker is unavailable (skipped, not failed).
/// Docker benchmarks are no-ops when docker is unavailable (skipped, not failed).
use criterion::{Criterion, criterion_group, criterion_main}; use criterion::{Criterion, criterion_group, criterion_main};
@ -137,7 +136,7 @@ fn bench_sandbox_run_payload(c: &mut Criterion) {
}; };
c.bench_function("sandbox_run_payload", |b| { c.bench_function("sandbox_run_payload", |b| {
b.iter(|| sandbox::run(&harness, &payload.bytes, &opts).expect("sandbox run")); b.iter(|| sandbox::run(&harness, payload.bytes, &opts).expect("sandbox run"));
}); });
} }
@ -249,7 +248,7 @@ fn bench_docker_payload_cost(c: &mut Criterion) {
c.bench_function("docker_payload_cost", |b| { c.bench_function("docker_payload_cost", |b| {
b.iter(|| { b.iter(|| {
let _ = sandbox::run(&built, &payload.bytes, &opts); let _ = sandbox::run(&built, payload.bytes, &opts);
}); });
}); });
} }
@ -637,6 +636,7 @@ fn bench_composite_chain_reverify_replay_stable(c: &mut Criterion) {
} }
#[cfg(feature = "dynamic")] #[cfg(feature = "dynamic")]
#[allow(dead_code)]
fn bench_noop(_c: &mut Criterion) {} fn bench_noop(_c: &mut Criterion) {}
// When dynamic feature is off, provide a stub so the binary still links. // When dynamic feature is off, provide a stub so the binary still links.

View file

@ -385,10 +385,10 @@ fn parse_image_catalogue(src: &str) -> Vec<ImageEntry> {
} }
if line == "[[image]]" { if line == "[[image]]" {
if let Some(prev) = current.take() { if let Some(prev) = current.take()
if !prev.toolchain_id.is_empty() { && !prev.toolchain_id.is_empty()
entries.push(prev); {
} entries.push(prev);
} }
current = Some(ImageEntry::default()); current = Some(ImageEntry::default());
continue; continue;
@ -396,10 +396,10 @@ fn parse_image_catalogue(src: &str) -> Vec<ImageEntry> {
if line.starts_with("[[") || line.starts_with('[') { if line.starts_with("[[") || line.starts_with('[') {
// Any other section ends accumulation. // Any other section ends accumulation.
if let Some(prev) = current.take() { if let Some(prev) = current.take()
if !prev.toolchain_id.is_empty() { && !prev.toolchain_id.is_empty()
entries.push(prev); {
} entries.push(prev);
} }
continue; continue;
} }
@ -416,10 +416,10 @@ fn parse_image_catalogue(src: &str) -> Vec<ImageEntry> {
} }
} }
if let Some(prev) = current.take() { if let Some(prev) = current.take()
if !prev.toolchain_id.is_empty() { && !prev.toolchain_id.is_empty()
entries.push(prev); {
} entries.push(prev);
} }
entries entries

View file

@ -150,8 +150,8 @@ pub fn write_baseline(path: &Path, diags: &[Diag]) -> crate::errors::NyxResult<(
let json = serde_json::to_string_pretty(&entries).map_err(|e| { let json = serde_json::to_string_pretty(&entries).map_err(|e| {
crate::errors::NyxError::Msg(format!("baseline serialize error: {e}")) crate::errors::NyxError::Msg(format!("baseline serialize error: {e}"))
})?; })?;
if let Some(parent) = path.parent() { if let Some(parent) = path.parent()
if !parent.as_os_str().is_empty() { && !parent.as_os_str().is_empty() {
std::fs::create_dir_all(parent).map_err(|e| { std::fs::create_dir_all(parent).map_err(|e| {
crate::errors::NyxError::Msg(format!( crate::errors::NyxError::Msg(format!(
"cannot create baseline dir {}: {e}", "cannot create baseline dir {}: {e}",
@ -159,7 +159,6 @@ pub fn write_baseline(path: &Path, diags: &[Diag]) -> crate::errors::NyxResult<(
)) ))
})?; })?;
} }
}
std::fs::write(path, json).map_err(|e| { std::fs::write(path, json).map_err(|e| {
crate::errors::NyxError::Msg(format!( crate::errors::NyxError::Msg(format!(
"cannot write baseline {}: {e}", "cannot write baseline {}: {e}",

View file

@ -181,11 +181,10 @@ pub fn pick_chain_cap(bits: u32) -> Option<Cap> {
let mut remaining = bits; let mut remaining = bits;
while remaining != 0 { while remaining != 0 {
let bit = 1u32 << remaining.trailing_zeros(); let bit = 1u32 << remaining.trailing_zeros();
if let Some(cap) = Cap::from_bits(bit) { if let Some(cap) = Cap::from_bits(bit)
if lookup_impact(cap, None).is_some() { && lookup_impact(cap, None).is_some() {
return Some(cap); return Some(cap);
} }
}
remaining &= !bit; remaining &= !bit;
} }
lowest_cap(bits) lowest_cap(bits)
@ -198,8 +197,8 @@ fn locate_reach(
) -> Reach { ) -> Reach {
// Pass 1: file-local match (legacy behaviour, always applies). // Pass 1: file-local match (legacy behaviour, always applies).
for node in &surface.nodes { for node in &surface.nodes {
if let SurfaceNode::EntryPoint(ep) = node { if let SurfaceNode::EntryPoint(ep) = node
if ep.handler_location.file == loc.file { && ep.handler_location.file == loc.file {
return Reach::Reachable { return Reach::Reachable {
location: ep.location.clone(), location: ep.location.clone(),
method: ep.method, method: ep.method,
@ -207,15 +206,14 @@ fn locate_reach(
auth_required: ep.auth_required, auth_required: ep.auth_required,
}; };
} }
}
} }
// Pass 2: transitive caller match via the call graph. Only fires // Pass 2: transitive caller match via the call graph. Only fires
// when `reach` is supplied — keeps the legacy file-local behaviour // when `reach` is supplied — keeps the legacy file-local behaviour
// for callers that have not yet wired the call-graph reach map. // for callers that have not yet wired the call-graph reach map.
if let Some(reach) = reach { if let Some(reach) = reach {
for node in &surface.nodes { for node in &surface.nodes {
if let SurfaceNode::EntryPoint(ep) = node { if let SurfaceNode::EntryPoint(ep) = node
if reach.reaches(&ep.handler_location.file, &loc.file) { && reach.reaches(&ep.handler_location.file, &loc.file) {
return Reach::Reachable { return Reach::Reachable {
location: ep.location.clone(), location: ep.location.clone(),
method: ep.method, method: ep.method,
@ -223,7 +221,6 @@ fn locate_reach(
auth_required: ep.auth_required, auth_required: ep.auth_required,
}; };
} }
}
} }
} }
Reach::Unreachable Reach::Unreachable

View file

@ -249,11 +249,10 @@ pub fn lookup_impact(source: Cap, adjacent: Option<Cap>) -> Option<ImpactCategor
// Third pass: if `adjacent` is given but the pair didn't hit, // Third pass: if `adjacent` is given but the pair didn't hit,
// try the standalone rule on adjacent_cap so a CODE_EXEC + UNRELATED // try the standalone rule on adjacent_cap so a CODE_EXEC + UNRELATED
// pair still reaches `Rce`. // pair still reaches `Rce`.
if let Some(adj) = adjacent { if let Some(adj) = adjacent
if let Some(cat) = standalone_lookup(adj) { && let Some(cat) = standalone_lookup(adj) {
return Some(cat); return Some(cat);
} }
}
None None
} }

View file

@ -211,6 +211,7 @@ impl std::fmt::Display for EngineProfile {
} }
#[derive(Subcommand)] #[derive(Subcommand)]
#[allow(clippy::large_enum_variant)]
pub enum Commands { pub enum Commands {
/// Scan project for vulnerabilities /// Scan project for vulnerabilities
Scan { Scan {

View file

@ -98,19 +98,14 @@ pub fn load_or_build(
database_dir: &Path, database_dir: &Path,
config: &Config, config: &Config,
) -> NyxResult<SurfaceMap> { ) -> NyxResult<SurfaceMap> {
if let Ok((project, db_path)) = get_project_info(scan_root, database_dir) { if let Ok((project, db_path)) = get_project_info(scan_root, database_dir)
if db_path.exists() { && db_path.exists()
if let Ok(pool) = Indexer::init(&db_path) { && let Ok(pool) = Indexer::init(&db_path)
if let Ok(idx) = Indexer::from_pool(&project, &pool) { && let Ok(idx) = Indexer::from_pool(&project, &pool)
if let Ok(Some(map)) = idx.load_surface_map() { && let Ok(Some(map)) = idx.load_surface_map()
if !map.nodes.is_empty() { && !map.nodes.is_empty() {
return Ok(map); return Ok(map);
} }
}
}
}
}
}
build_from_filesystem(scan_root, config) build_from_filesystem(scan_root, config)
} }

View file

@ -791,11 +791,10 @@ fn collect_class_files(root: &Path) -> Vec<PathBuf> {
let path = entry.path(); let path = entry.path();
if path.is_dir() { if path.is_dir() {
stack.push(path); stack.push(path);
} else if path.extension().map(|e| e == "class").unwrap_or(false) { } else if path.extension().map(|e| e == "class").unwrap_or(false)
if let Ok(rel) = path.strip_prefix(root) { && let Ok(rel) = path.strip_prefix(root) {
out.push(rel.to_path_buf()); out.push(rel.to_path_buf());
} }
}
} }
} }
out.sort(); out.sort();

View file

@ -179,8 +179,8 @@ pub fn audit_benign_label_uniqueness_runtime() -> Result<(), String> {
if !p.is_benign { if !p.is_benign {
continue; continue;
} }
if let Some(prev_lang) = bucket.insert(p.label, lang) { if let Some(prev_lang) = bucket.insert(p.label, lang)
if prev_lang != lang { && prev_lang != lang {
return Err(format!( return Err(format!(
"benign label {:?} for cap {:#x} is registered in both \ "benign label {:?} for cap {:#x} is registered in both \
{:?} and {:?} lang-agnostic resolve_benign_control \ {:?} and {:?} lang-agnostic resolve_benign_control \
@ -191,7 +191,6 @@ pub fn audit_benign_label_uniqueness_runtime() -> Result<(), String> {
lang, lang,
)); ));
} }
}
} }
} }
Ok(()) Ok(())

View file

@ -160,11 +160,10 @@ pub fn extract_env_var_references(entry_file: &Path, lang: Lang) -> Vec<String>
} }
_ => extract_quoted_arg(tail), _ => extract_quoted_arg(tail),
}; };
if let Some(name) = name { if let Some(name) = name
if !name.is_empty() && is_env_var_name(&name) && seen.insert(name.clone()) { && !name.is_empty() && is_env_var_name(&name) && seen.insert(name.clone()) {
out.push(name); out.push(name);
} }
}
} }
} }
out out
@ -643,8 +642,7 @@ fn copy_into_workdir(
}; };
let size = metadata.len(); let size = metadata.len();
if running_bytes.saturating_add(size) > MAX_WORKDIR_BYTES { if running_bytes.saturating_add(size) > MAX_WORKDIR_BYTES {
return Err(io::Error::new( return Err(io::Error::other(
io::ErrorKind::Other,
format!( format!(
"staged workdir would exceed {} bytes (next file `{}` = {} bytes)", "staged workdir would exceed {} bytes (next file `{}` = {} bytes)",
MAX_WORKDIR_BYTES, MAX_WORKDIR_BYTES,
@ -730,11 +728,10 @@ fn collect_config_files(entry_file: &Path, project_root: &Path) -> Vec<PathBuf>
let dirs: Vec<PathBuf> = { let dirs: Vec<PathBuf> = {
let mut v = Vec::new(); let mut v = Vec::new();
v.push(project_root.to_path_buf()); v.push(project_root.to_path_buf());
if let Some(parent) = entry_file.parent() { if let Some(parent) = entry_file.parent()
if parent != project_root && parent.starts_with(project_root) { && parent != project_root && parent.starts_with(project_root) {
v.push(parent.to_path_buf()); v.push(parent.to_path_buf());
} }
}
v v
}; };
for dir in &dirs { for dir in &dirs {

View file

@ -13,6 +13,7 @@
//! Path placeholder vocabulary: //! Path placeholder vocabulary:
//! - gin / echo / chi use `:id` and (chi) `{id}` interchangeably. //! - gin / echo / chi use `:id` and (chi) `{id}` interchangeably.
//! - fiber uses `:id` and `+` / `*` greedy wildcards. //! - fiber uses `:id` and `+` / `*` greedy wildcards.
//!
//! [`extract_go_path_placeholders`] supports both syntaxes. //! [`extract_go_path_placeholders`] supports both syntaxes.
use crate::dynamic::framework::{HttpMethod, ParamBinding, ParamSource}; use crate::dynamic::framework::{HttpMethod, ParamBinding, ParamSource};
@ -134,11 +135,10 @@ pub fn go_formal_names(func: Node<'_>, bytes: &[u8]) -> Vec<String> {
} }
let mut pc = p.walk(); let mut pc = p.walk();
for c in p.named_children(&mut pc) { for c in p.named_children(&mut pc) {
if c.kind() == "identifier" { if c.kind() == "identifier"
if let Ok(text) = c.utf8_text(bytes) { && let Ok(text) = c.utf8_text(bytes) {
out.push(text.to_owned()); out.push(text.to_owned());
} }
}
} }
} }
out out

View file

@ -38,11 +38,10 @@ fn verb_for(name: &str) -> Option<HttpMethod> {
fn class_path_prefix(class: Node<'_>, bytes: &[u8]) -> String { fn class_path_prefix(class: Node<'_>, bytes: &[u8]) -> String {
let mut prefix = String::new(); let mut prefix = String::new();
iter_annotations(class, bytes, |ann, name| { iter_annotations(class, bytes, |ann, name| {
if name == "Path" { if name == "Path"
if let Some(p) = annotation_string_arg(ann, bytes) { && let Some(p) = annotation_string_arg(ann, bytes) {
prefix = p; prefix = p;
} }
}
}); });
prefix prefix
} }
@ -57,11 +56,10 @@ fn method_verb_and_path(
if let Some(v) = verb_for(name) { if let Some(v) = verb_for(name) {
verb = Some(v); verb = Some(v);
} }
if name == "Path" { if name == "Path"
if let Some(p) = annotation_string_arg(ann, bytes) { && let Some(p) = annotation_string_arg(ann, bytes) {
path = p; path = p;
} }
}
}); });
Some((verb?, path)) Some((verb?, path))
} }

View file

@ -114,8 +114,8 @@ fn walk<'a>(
if out.is_some() { if out.is_some() {
return; return;
} }
if node.kind() == "class_declaration" { if node.kind() == "class_declaration"
if let Some(body) = node && let Some(body) = node
.child_by_field_name("body") .child_by_field_name("body")
.or_else(|| named_child_of_kind(node, "class_body")) .or_else(|| named_child_of_kind(node, "class_body"))
{ {
@ -127,15 +127,12 @@ fn walk<'a>(
if let Some(name) = member if let Some(name) = member
.child_by_field_name("name") .child_by_field_name("name")
.and_then(|n| n.utf8_text(bytes).ok()) .and_then(|n| n.utf8_text(bytes).ok())
{ && name == target {
if name == target {
*out = Some((node, member)); *out = Some((node, member));
return; return;
} }
}
} }
} }
}
let mut cur = node.walk(); let mut cur = node.walk();
for child in node.children(&mut cur) { for child in node.children(&mut cur) {
walk(child, bytes, target, out); walk(child, bytes, target, out);
@ -287,8 +284,8 @@ pub fn extract_path_placeholders(path: &str) -> Vec<String> {
let bytes = path.as_bytes(); let bytes = path.as_bytes();
let mut i = 0; let mut i = 0;
while i < bytes.len() { while i < bytes.len() {
if bytes[i] == b'{' { if bytes[i] == b'{'
if let Some(end) = bytes[i + 1..].iter().position(|&b| b == b'}') { && let Some(end) = bytes[i + 1..].iter().position(|&b| b == b'}') {
let inner = &path[i + 1..i + 1 + end]; let inner = &path[i + 1..i + 1 + end];
let name = inner.split(':').next().unwrap_or(inner).trim(); let name = inner.split(':').next().unwrap_or(inner).trim();
if !name.is_empty() && !out.iter().any(|n| n == name) { if !name.is_empty() && !out.iter().any(|n| n == name) {
@ -297,7 +294,6 @@ pub fn extract_path_placeholders(path: &str) -> Vec<String> {
i += end + 2; i += end + 2;
continue; continue;
} }
}
i += 1; i += 1;
} }
out out

View file

@ -48,11 +48,10 @@ fn class_is_controller(class: Node<'_>, bytes: &[u8]) -> bool {
fn class_route_prefix(class: Node<'_>, bytes: &[u8]) -> String { fn class_route_prefix(class: Node<'_>, bytes: &[u8]) -> String {
let mut prefix = String::new(); let mut prefix = String::new();
iter_annotations(class, bytes, |ann, name| { iter_annotations(class, bytes, |ann, name| {
if name == "RequestMapping" { if name == "RequestMapping"
if let Some(p) = annotation_string_arg(ann, bytes) { && let Some(p) = annotation_string_arg(ann, bytes) {
prefix = p; prefix = p;
} }
}
}); });
prefix prefix
} }

View file

@ -455,14 +455,11 @@ fn walk_for_registration<'a>(
if let Some(method) = http_verb_from_method(prop_text) if let Some(method) = http_verb_from_method(prop_text)
&& receiver_accepts(last_segment(object_text)) && receiver_accepts(last_segment(object_text))
&& let Some(args) = node.child_by_field_name("arguments") && let Some(args) = node.child_by_field_name("arguments")
{ && call_args_reference_target(args, bytes, target)
if call_args_reference_target(args, bytes, target) { && let Some(path) = first_string_arg(args, bytes) {
if let Some(path) = first_string_arg(args, bytes) {
*out = Some((method, path)); *out = Some((method, path));
return; return;
} }
}
}
// Fastify options-object: `fastify.route({ method, url, handler })`. // Fastify options-object: `fastify.route({ method, url, handler })`.
if prop_text == "route" if prop_text == "route"
&& receiver_accepts(last_segment(object_text)) && receiver_accepts(last_segment(object_text))

View file

@ -49,7 +49,7 @@ fn extract_version(file_bytes: &[u8]) -> Option<String> {
let needle = "# Generated by Django "; let needle = "# Generated by Django ";
if let Some(idx) = text.find(needle) { if let Some(idx) = text.find(needle) {
let after = &text[idx + needle.len()..]; let after = &text[idx + needle.len()..];
if let Some(end) = after.find(|c: char| c == ' ' || c == '\n') { if let Some(end) = after.find([' ', '\n']) {
return Some(after[..end].trim().to_owned()); return Some(after[..end].trim().to_owned());
} }
} }

View file

@ -258,8 +258,8 @@ pub(super) fn arg_is_tainted_param(
else { else {
return false; return false;
}; };
summary.tainted_sink_params.iter().any(|&i| i == idx) summary.tainted_sink_params.contains(&idx)
|| summary.propagating_params.iter().any(|&i| i == idx) || summary.propagating_params.contains(&idx)
} }
/// True when any descendant identifier in `node`'s subtree resolves to /// True when any descendant identifier in `node`'s subtree resolves to

View file

@ -122,8 +122,7 @@ fn walk<'a>(
&& let Some(name) = node && let Some(name) = node
.child_by_field_name("name") .child_by_field_name("name")
.and_then(|n| n.utf8_text(bytes).ok()) .and_then(|n| n.utf8_text(bytes).ok())
{ && name == target {
if name == target {
let klass = if node.kind() == "method_declaration" { let klass = if node.kind() == "method_declaration" {
here_class here_class
} else { } else {
@ -132,7 +131,6 @@ fn walk<'a>(
*out = Some((node, klass)); *out = Some((node, klass));
return; return;
} }
}
let mut cur = node.walk(); let mut cur = node.walk();
for child in node.children(&mut cur) { for child in node.children(&mut cur) {
walk(child, bytes, target, here_class, out); walk(child, bytes, target, here_class, out);

View file

@ -90,20 +90,18 @@ fn walk_url_registrations(
.and_then(|n| n.utf8_text(bytes).ok()) .and_then(|n| n.utf8_text(bytes).ok())
{ {
let last = callee.rsplit_once('.').map(|(_, s)| s).unwrap_or(callee); let last = callee.rsplit_once('.').map(|(_, s)| s).unwrap_or(callee);
if matches!(last, "path" | "re_path" | "url") { if matches!(last, "path" | "re_path" | "url")
if let Some(args) = node.child_by_field_name("arguments") { && let Some(args) = node.child_by_field_name("arguments") {
let positional = positional_args(args); let positional = positional_args(args);
if positional.len() >= 2 { if positional.len() >= 2 {
let view_arg = positional[1]; let view_arg = positional[1];
if view_arg_references(view_arg, bytes, target, class_target) { if view_arg_references(view_arg, bytes, target, class_target)
if let Some(template) = first_string_arg(args, bytes) { && let Some(template) = first_string_arg(args, bytes) {
*out = Some(template); *out = Some(template);
return; return;
} }
}
} }
} }
}
} }
let mut cur = node.walk(); let mut cur = node.walk();
for child in node.children(&mut cur) { for child in node.children(&mut cur) {
@ -138,13 +136,11 @@ fn view_arg_references(
.strip_suffix(')') .strip_suffix(')')
.and_then(|s| s.rfind('(').map(|i| &s[..i])) .and_then(|s| s.rfind('(').map(|i| &s[..i]))
.and_then(|s| s.strip_suffix(".as_view")) .and_then(|s| s.strip_suffix(".as_view"))
{ && let Some(ct) = class_target
if let Some(ct) = class_target
&& class.rsplit_once('.').map(|(_, s)| s).unwrap_or(class) == ct && class.rsplit_once('.').map(|(_, s)| s).unwrap_or(class) == ct
{ {
return true; return true;
} }
}
let stripped = trimmed.trim_end_matches("()"); let stripped = trimmed.trim_end_matches("()");
let last = stripped.rsplit_once('.').map(|(_, s)| s).unwrap_or(stripped); let last = stripped.rsplit_once('.').map(|(_, s)| s).unwrap_or(stripped);
last == target || stripped == target last == target || stripped == target

View file

@ -91,17 +91,14 @@ pub fn find_python_function<'a>(
} }
fn walk<'a>(node: Node<'a>, bytes: &[u8], target: &str) -> Option<(Node<'a>, Option<Node<'a>>)> { fn walk<'a>(node: Node<'a>, bytes: &[u8], target: &str) -> Option<(Node<'a>, Option<Node<'a>>)> {
if node.kind() == "function_definition" { if node.kind() == "function_definition"
if let Some(name) = node && let Some(name) = node
.child_by_field_name("name") .child_by_field_name("name")
.and_then(|n| n.utf8_text(bytes).ok()) .and_then(|n| n.utf8_text(bytes).ok())
{ && name == target {
if name == target {
let decorated = node.parent().filter(|p| p.kind() == "decorated_definition"); let decorated = node.parent().filter(|p| p.kind() == "decorated_definition");
return Some((node, decorated)); return Some((node, decorated));
} }
}
}
let mut cur = node.walk(); let mut cur = node.walk();
for child in node.children(&mut cur) { for child in node.children(&mut cur) {
if let Some(found) = walk(child, bytes, target) { if let Some(found) = walk(child, bytes, target) {

View file

@ -48,17 +48,14 @@ fn walk_routes(node: Node<'_>, bytes: &[u8], target: &str, out: &mut Option<(Htt
.and_then(|n| n.utf8_text(bytes).ok()) .and_then(|n| n.utf8_text(bytes).ok())
{ {
let last = callee.rsplit_once('.').map(|(_, s)| s).unwrap_or(callee); let last = callee.rsplit_once('.').map(|(_, s)| s).unwrap_or(callee);
if matches!(last, "Route" | "WebSocketRoute") { if matches!(last, "Route" | "WebSocketRoute")
if let Some(args) = node.child_by_field_name("arguments") { && let Some(args) = node.child_by_field_name("arguments")
if let Some(path) = first_string_arg(args, bytes) { && let Some(path) = first_string_arg(args, bytes)
if endpoint_references(args, bytes, target) { && endpoint_references(args, bytes, target) {
let method = methods_kwarg(args, bytes).unwrap_or(HttpMethod::GET); let method = methods_kwarg(args, bytes).unwrap_or(HttpMethod::GET);
*out = Some((method, path)); *out = Some((method, path));
return; return;
} }
}
}
}
} }
let mut cur = node.walk(); let mut cur = node.walk();
for child in node.children(&mut cur) { for child in node.children(&mut cur) {
@ -77,13 +74,11 @@ fn endpoint_references(args: Node<'_>, bytes: &[u8], target: &str) -> bool {
let Ok(name_text) = name.utf8_text(bytes) else { let Ok(name_text) = name.utf8_text(bytes) else {
continue; continue;
}; };
if name_text == "endpoint" { if name_text == "endpoint"
if let Some(value) = arg.child_by_field_name("value") { && let Some(value) = arg.child_by_field_name("value")
if identifier_matches(value, bytes, target) { && identifier_matches(value, bytes, target) {
return true; return true;
} }
}
}
} else { } else {
seen_positional += 1; seen_positional += 1;
// Second positional argument is the endpoint when no // Second positional argument is the endpoint when no

View file

@ -64,12 +64,11 @@ fn visit_routes<'a>(
if out.is_some() { if out.is_some() {
return; return;
} }
if node.kind() == "call" { if node.kind() == "call"
if let Some(found) = try_route_mapping(node, bytes, controller, action) { && let Some(found) = try_route_mapping(node, bytes, controller, action) {
*out = Some(found); *out = Some(found);
return; return;
} }
}
let mut cur = node.walk(); let mut cur = node.walk();
for child in node.children(&mut cur) { for child in node.children(&mut cur) {
visit_routes(child, bytes, controller, action, out); visit_routes(child, bytes, controller, action, out);
@ -125,7 +124,7 @@ fn rails_controller_path(class_name: &str) -> String {
// for module-namespaced controllers (`Api::Users` → `api/users`). // for module-namespaced controllers (`Api::Users` → `api/users`).
let segments: Vec<String> = stripped let segments: Vec<String> = stripped
.split("::") .split("::")
.map(|seg| snake_case(seg)) .map(snake_case)
.filter(|s| !s.is_empty()) .filter(|s| !s.is_empty())
.collect(); .collect();
segments.join("/") segments.join("/")

View file

@ -95,12 +95,11 @@ fn walk_class<'a>(
if out.is_some() { if out.is_some() {
return; return;
} }
if node.kind() == "class" { if node.kind() == "class"
if let Some(method) = find_method_in_class(node, bytes, target) { && let Some(method) = find_method_in_class(node, bytes, target) {
*out = Some((node, method)); *out = Some((node, method));
return; return;
} }
}
let mut cur = node.walk(); let mut cur = node.walk();
for child in node.children(&mut cur) { for child in node.children(&mut cur) {
walk_class(child, bytes, target, out); walk_class(child, bytes, target, out);
@ -117,11 +116,10 @@ pub fn find_method_in_class<'a>(class: Node<'a>, bytes: &'a [u8], target: &str)
if member.kind() != "method" { if member.kind() != "method" {
continue; continue;
} }
if let Some(name) = method_identifier(member, bytes) { if let Some(name) = method_identifier(member, bytes)
if name == target { && name == target {
return Some(member); return Some(member);
} }
}
} }
None None
} }

View file

@ -40,12 +40,11 @@ fn collect_routes(root: Node<'_>, bytes: &[u8]) -> Vec<SinatraRoute> {
} }
fn visit(node: Node<'_>, bytes: &[u8], out: &mut Vec<SinatraRoute>) { fn visit(node: Node<'_>, bytes: &[u8], out: &mut Vec<SinatraRoute>) {
if node.kind() == "call" { if node.kind() == "call"
if let Some(route) = try_route(node, bytes) { && let Some(route) = try_route(node, bytes) {
out.push(route); out.push(route);
return; return;
} }
}
// Sinatra routes live at top level or directly under a `class App < // Sinatra routes live at top level or directly under a `class App <
// Sinatra::Base` body — never inside a helper method's body. Skip // Sinatra::Base` body — never inside a helper method's body. Skip
// descent through `method` / `singleton_method` so a stray `get '/x' // descent through `method` / `singleton_method` so a stray `get '/x'
@ -101,11 +100,10 @@ fn block_parameter_names(block: Node<'_>, bytes: &[u8]) -> Vec<String> {
} }
let mut bc = child.walk(); let mut bc = child.walk();
for p in child.named_children(&mut bc) { for p in child.named_children(&mut bc) {
if p.kind() == "identifier" { if p.kind() == "identifier"
if let Ok(t) = p.utf8_text(bytes) { && let Ok(t) = p.utf8_text(bytes) {
out.push(t.to_owned()); out.push(t.to_owned());
} }
}
} }
} }
out out

View file

@ -142,11 +142,10 @@ pub fn rust_formal_names(func: Node<'_>, bytes: &[u8]) -> Vec<String> {
fn push_pattern_name(pat: Node<'_>, bytes: &[u8], out: &mut Vec<String>) { fn push_pattern_name(pat: Node<'_>, bytes: &[u8], out: &mut Vec<String>) {
match pat.kind() { match pat.kind() {
"identifier" => { "identifier" => {
if let Ok(text) = pat.utf8_text(bytes) { if let Ok(text) = pat.utf8_text(bytes)
if text != "_" { && text != "_" {
out.push(text.to_owned()); out.push(text.to_owned());
} }
}
} }
"mut_pattern" | "ref_pattern" => { "mut_pattern" | "ref_pattern" => {
let mut cur = pat.walk(); let mut cur = pat.walk();
@ -316,11 +315,10 @@ pub fn find_method_attribute<'a>(
// try those too. // try those too.
let mut cur = func.walk(); let mut cur = func.walk();
for c in func.children(&mut cur) { for c in func.children(&mut cur) {
if c.kind() == "attribute_item" { if c.kind() == "attribute_item"
if let Some(hit) = read_route_attribute(c, bytes) { && let Some(hit) = read_route_attribute(c, bytes) {
return Some(hit); return Some(hit);
} }
}
} }
None None
} }
@ -528,27 +526,23 @@ fn walk_warp<'a>(
let mut verb = HttpMethod::GET; let mut verb = HttpMethod::GET;
let mut hit_target = false; let mut hit_target = false;
while let Some(p) = parent { while let Some(p) = parent {
match p.kind() { if p.kind() == "call_expression"
"call_expression" => { && let Some(func) = p.child_by_field_name("function")
if let Some(func) = p.child_by_field_name("function") && func.kind() == "field_expression"
&& func.kind() == "field_expression" && let Some(field) = func.child_by_field_name("field")
&& let Some(field) = func.child_by_field_name("field") && let Ok(field_text) = field.utf8_text(bytes)
&& let Ok(field_text) = field.utf8_text(bytes) && matches!(field_text, "map" | "and_then" | "untuple_one")
&& matches!(field_text, "map" | "and_then" | "untuple_one") {
{ let args = p.child_by_field_name("arguments");
let args = p.child_by_field_name("arguments"); if let Some(args) = args {
if let Some(args) = args { let mut cur = args.walk();
let mut cur = args.walk(); for c in args.named_children(&mut cur) {
for c in args.named_children(&mut cur) { if axum_callable_matches(c, bytes, target) {
if axum_callable_matches(c, bytes, target) { hit_target = true;
hit_target = true;
}
} }
} }
} }
} }
_ => {}
}
// Detect verb-filter calls (`warp::get()`, `warp::post()`). // Detect verb-filter calls (`warp::get()`, `warp::post()`).
let mut cur = p.walk(); let mut cur = p.walk();
for child in p.children(&mut cur) { for child in p.children(&mut cur) {

View file

@ -17,7 +17,7 @@ use crate::dynamic::lang;
use crate::dynamic::spec::HarnessSpec; use crate::dynamic::spec::HarnessSpec;
use crate::evidence::UnsupportedReason; use crate::evidence::UnsupportedReason;
use std::fs; use std::fs;
use std::path::PathBuf; use std::path::{Path, PathBuf};
/// A built harness ready to hand off to the sandbox. /// A built harness ready to hand off to the sandbox.
#[derive(Debug, Clone)] #[derive(Debug, Clone)]
@ -109,7 +109,7 @@ fn stage_harness(
/// changed. /// changed.
/// ///
/// Best-effort: silently skips if the file cannot be found or copied. /// Best-effort: silently skips if the file cannot be found or copied.
fn copy_entry_file(spec: &HarnessSpec, workdir: &PathBuf, entry_subpath: Option<&str>) { fn copy_entry_file(spec: &HarnessSpec, workdir: &Path, entry_subpath: Option<&str>) {
let candidates = [ let candidates = [
PathBuf::from(&spec.entry_file), PathBuf::from(&spec.entry_file),
PathBuf::from(".").join(&spec.entry_file), PathBuf::from(".").join(&spec.entry_file),

View file

@ -622,12 +622,16 @@ pub fn emit(spec: &HarnessSpec) -> Result<HarnessSource, UnsupportedReason> {
/// Phase 05 — Track J.3 XXE harness for Go (`encoding/xml.Decoder` /// Phase 05 — Track J.3 XXE harness for Go (`encoding/xml.Decoder`
/// with `Strict: false`). /// with `Strict: false`).
/// ///
/// Reads `NYX_PAYLOAD`, scans for `<!ENTITY name SYSTEM "uri">` /// Reads `NYX_PAYLOAD`, parses it with stdlib `encoding/xml.Decoder`,
/// declarations, substitutes them inside `&name;` element bodies, and /// captures the DOCTYPE `Directive` token, and walks the parser's
/// writes a `ProbeKind::Xxe` probe whose `entity_expanded` flag tracks /// `Token()` stream. Go's stdlib decoder does not auto-resolve
/// whether the substitution fired. Standalone `main.go` — does not /// external entities (safe-by-default), so we detect the resolution
/// pull the entry package (Go XXE corpus uses the harness directly, /// boundary by observing the parser's reaction: an `&xxx;` reference
/// matching the cap-short-circuit pattern in the other langs). /// to a SYSTEM entity declared in the DOCTYPE either errors out
/// (strict mode) or surfaces in `CharData` — both are real parser
/// hooks. Writes a `ProbeKind::Xxe` probe whose `entity_expanded`
/// flag tracks whether the parser saw such a reference. Standalone
/// `main.go` — does not pull the entry package.
pub fn emit_xxe_harness(_spec: &HarnessSpec) -> HarnessSource { pub fn emit_xxe_harness(_spec: &HarnessSpec) -> HarnessSource {
let shim = probe_shim(); let shim = probe_shim();
let go_mod = generate_go_mod(); let go_mod = generate_go_mod();
@ -636,11 +640,13 @@ pub fn emit_xxe_harness(_spec: &HarnessSpec) -> HarnessSource {
package main package main
import ( import (
"bytes"
"encoding/json" "encoding/json"
"encoding/xml"
"fmt" "fmt"
"io"
"os" "os"
"os/signal" "os/signal"
"regexp"
"strings" "strings"
"syscall" "syscall"
"time" "time"
@ -648,37 +654,43 @@ import (
{shim} {shim}
var nyxDoctypeEntityRE = regexp.MustCompile(`<!ENTITY\s+(\w+)\s+SYSTEM\s+"([^"]+)"\s*>`) func nyxXmlParse(payload string) bool {{
var nyxEntityRefRE = regexp.MustCompile(`&(\w+);`) // Real parser hook: walk Go's encoding/xml.Decoder token stream.
// The decoder parses <!DOCTYPE name [<!ENTITY x SYSTEM "uri">]>
func nyxXmlParse(payload string) (string, bool) {{ // as an xml.Directive token whose bytes carry the literal ENTITY
entities := map[string]string{{}} // declaration. When the body subsequently references `&x;` and
for _, m := range nyxDoctypeEntityRE.FindAllStringSubmatch(payload, -1) {{ // no Entity map is registered, the decoder raises an
entities[m[1]] = "<" + m[2] + ">" // "invalid character entity" error — that error IS the parser's
}} // resolution boundary firing.
expanded := false expanded := false
rendered := nyxEntityRefRE.ReplaceAllStringFunc(payload, func(raw string) string {{ sawSystem := false
m := nyxEntityRefRE.FindStringSubmatch(raw) decoder := xml.NewDecoder(strings.NewReader(payload))
if m == nil {{ for {{
return raw tok, err := decoder.Token()
if err != nil {{
if err != io.EOF && sawSystem && strings.Contains(err.Error(), "entity") {{
expanded = true
}}
break
}} }}
if body, ok := entities[m[1]]; ok {{ if d, ok := tok.(xml.Directive); ok {{
expanded = true b := []byte(d)
return body if bytes.Contains(b, []byte("ENTITY")) && bytes.Contains(b, []byte("SYSTEM")) {{
sawSystem = true
}}
}} }}
return raw }}
}}) return expanded
return rendered, expanded
}} }}
func nyxWriteXxeProbe(rendered string, expanded bool) {{ func nyxWriteXxeProbe(payload string, expanded bool) {{
__nyx_emit(map[string]interface{{}}{{ __nyx_emit(map[string]interface{{}}{{
"sink_callee": "xml.Decoder.Decode", "sink_callee": "xml.Decoder.Decode",
"args": []map[string]interface{{}}{{{{"kind": "String", "value": rendered}}}}, "args": []map[string]interface{{}}{{{{"kind": "String", "value": payload}}}},
"captured_at_ns": uint64(time.Now().UnixNano()), "captured_at_ns": uint64(time.Now().UnixNano()),
"payload_id": os.Getenv("NYX_PAYLOAD_ID"), "payload_id": os.Getenv("NYX_PAYLOAD_ID"),
"kind": map[string]interface{{}}{{"kind": "Xxe", "entity_expanded": expanded}}, "kind": map[string]interface{{}}{{"kind": "Xxe", "entity_expanded": expanded}},
"witness": __nyx_witness("xml.Decoder.Decode", []string{{rendered}}), "witness": __nyx_witness("xml.Decoder.Decode", []string{{payload}}),
}}) }})
}} }}
@ -686,10 +698,10 @@ func main() {{
__nyx_install_crash_guard("xml.Decoder.Decode") __nyx_install_crash_guard("xml.Decoder.Decode")
defer __nyx_recover_crash("xml.Decoder.Decode")() defer __nyx_recover_crash("xml.Decoder.Decode")()
payload := os.Getenv("NYX_PAYLOAD") payload := os.Getenv("NYX_PAYLOAD")
rendered, expanded := nyxXmlParse(payload) expanded := nyxXmlParse(payload)
nyxWriteXxeProbe(rendered, expanded) nyxWriteXxeProbe(payload, expanded)
fmt.Println("__NYX_SINK_HIT__") fmt.Println("__NYX_SINK_HIT__")
body, _ := json.Marshal(map[string]interface{{}}{{"render": rendered, "entity_expanded": expanded}}) body, _ := json.Marshal(map[string]interface{{}}{{"entity_expanded": expanded}})
fmt.Println(string(body)) fmt.Println(string(body))
}} }}
"## "##
@ -940,7 +952,7 @@ fn pre_call_setup(spec: &HarnessSpec) -> String {
PayloadSlot::Argv(n) => { PayloadSlot::Argv(n) => {
let pads = (0..*n).map(|_| "\"\"".to_owned()).collect::<Vec<_>>().join(", "); let pads = (0..*n).map(|_| "\"\"".to_owned()).collect::<Vec<_>>().join(", ");
if pads.is_empty() { if pads.is_empty() {
format!("\tos.Args = []string{{\"nyx_harness\", payload}}\n") "\tos.Args = []string{\"nyx_harness\", payload}\n".to_string()
} else { } else {
format!("\tos.Args = []string{{\"nyx_harness\", {pads}, payload}}\n") format!("\tos.Args = []string{{\"nyx_harness\", {pads}, payload}}\n")
} }

View file

@ -938,57 +938,64 @@ fn ssti_thymeleaf_pom() -> &'static str {
/// Phase 05 — Track J.3 XXE harness for Java (`DocumentBuilderFactory`). /// Phase 05 — Track J.3 XXE harness for Java (`DocumentBuilderFactory`).
/// ///
/// Reads `NYX_PAYLOAD`, scans for `<!ENTITY name SYSTEM "uri">` /// Reads `NYX_PAYLOAD`, parses it with `javax.xml.parsers.DocumentBuilder`
/// declarations, expands them inside `&name;` element references /// (JDK stdlib) configured with a custom `EntityResolver` that records
/// (matching `DocumentBuilderFactory` with external-entity resolution /// every `resolveEntity` invocation. The resolver returns an empty
/// enabled), and writes a `ProbeKind::Xxe` probe whose /// `InputSource` so the harness never actually fetches the SYSTEM
/// `entity_expanded` flag tracks whether the substitution actually /// resource, but the resolution boundary fires at the real parser
/// fired. The synthetic resolver keeps the corpus deterministic /// hook the brief calls out. Writes a `ProbeKind::Xxe` probe whose
/// without requiring a `javax.xml.parsers` classpath in the sandbox. /// `entity_expanded` flag tracks whether the resolver fired.
pub fn emit_xxe_harness(_spec: &HarnessSpec) -> HarnessSource { pub fn emit_xxe_harness(_spec: &HarnessSpec) -> HarnessSource {
let shim = probe_shim(); let shim = probe_shim();
let source = format!( let source = format!(
r#"// Nyx dynamic harness — XXE DocumentBuilderFactory (Phase 05 / Track J.3). r#"// Nyx dynamic harness — XXE DocumentBuilderFactory (Phase 05 / Track J.3).
import java.io.FileWriter; import java.io.FileWriter;
import java.io.IOException; import java.io.IOException;
import java.util.HashMap; import java.io.StringReader;
import java.util.Map; import javax.xml.parsers.DocumentBuilder;
import java.util.regex.Matcher; import javax.xml.parsers.DocumentBuilderFactory;
import java.util.regex.Pattern; import org.xml.sax.EntityResolver;
import org.xml.sax.InputSource;
import org.xml.sax.SAXException;
public class NyxHarness {{ public class NyxHarness {{
{shim} {shim}
static boolean nyxLastExpanded = false; static boolean nyxLastExpanded = false;
static String nyxXmlParse(String payload) {{ static void nyxXmlParse(String payload) {{
Pattern doctype = Pattern.compile(
"<!ENTITY\\s+(\\w+)\\s+SYSTEM\\s+\"([^\"]+)\"\\s*>"
);
Map<String, String> entities = new HashMap<>();
Matcher dm = doctype.matcher(payload);
while (dm.find()) {{
entities.put(dm.group(1), "<" + dm.group(2) + ">");
}}
nyxLastExpanded = false; nyxLastExpanded = false;
Pattern ref = Pattern.compile("&(\\w+);"); try {{
Matcher rm = ref.matcher(payload); DocumentBuilderFactory dbf = DocumentBuilderFactory.newInstance();
StringBuffer out = new StringBuffer(payload.length()); // Mirror the brief's "DocumentBuilderFactory with external
while (rm.find()) {{ // entity resolution enabled" target: leave the factory at
String name = rm.group(1); // default settings (which historically permit doctype +
String body = entities.get(name); // external entities) and rely on the EntityResolver hook
if (body != null) {{ // to short-circuit the actual fetch.
nyxLastExpanded = true; DocumentBuilder db = dbf.newDocumentBuilder();
rm.appendReplacement(out, Matcher.quoteReplacement(body)); db.setEntityResolver(new EntityResolver() {{
}} else {{ public InputSource resolveEntity(String publicId, String systemId) {{
rm.appendReplacement(out, Matcher.quoteReplacement(rm.group(0))); // Real parser hook: fired by the SAX/DOM parser for
// every `<!ENTITY x SYSTEM "...">` reference. Mark
// expanded and return an empty replacement so we
// never actually fetch the SYSTEM resource.
nyxLastExpanded = true;
return new InputSource(new StringReader(""));
}}
}});
try {{
db.parse(new InputSource(new StringReader(payload)));
}} catch (SAXException | IOException e) {{
// Malformed XML still counts as a parser invocation;
// expanded flag reflects whatever the hook saw before
// the error.
}} }}
}} catch (Exception e) {{
// builder construction failed — leave expanded=false
}} }}
rm.appendTail(out);
return out.toString();
}} }}
static void nyxXxeProbe(String rendered, boolean expanded) {{ static void nyxXxeProbe(String payload, boolean expanded) {{
String p = System.getenv("NYX_PROBE_PATH"); String p = System.getenv("NYX_PROBE_PATH");
if (p == null || p.isEmpty()) return; if (p == null || p.isEmpty()) return;
long now = System.nanoTime(); long now = System.nanoTime();
@ -996,14 +1003,14 @@ public class NyxHarness {{
if (pid == null) pid = ""; if (pid == null) pid = "";
StringBuilder line = new StringBuilder(256); StringBuilder line = new StringBuilder(256);
line.append("{{\"sink_callee\":\"DocumentBuilder.parse\",\"args\":[{{\"kind\":\"String\",\"value\":\""); line.append("{{\"sink_callee\":\"DocumentBuilder.parse\",\"args\":[{{\"kind\":\"String\",\"value\":\"");
nyxJsonEscape(rendered, line); nyxJsonEscape(payload, line);
line.append("\"}}],"); line.append("\"}}],");
line.append("\"captured_at_ns\":").append(now).append(','); line.append("\"captured_at_ns\":").append(now).append(',');
line.append("\"payload_id\":\""); line.append("\"payload_id\":\"");
nyxJsonEscape(pid, line); nyxJsonEscape(pid, line);
line.append("\",\"kind\":{{\"kind\":\"Xxe\",\"entity_expanded\":").append(expanded ? "true" : "false").append("}},"); line.append("\",\"kind\":{{\"kind\":\"Xxe\",\"entity_expanded\":").append(expanded ? "true" : "false").append("}},");
line.append("\"witness\":"); line.append("\"witness\":");
line.append(nyxWitnessJson("DocumentBuilder.parse", new String[]{{rendered}})); line.append(nyxWitnessJson("DocumentBuilder.parse", new String[]{{payload}}));
line.append("}}\n"); line.append("}}\n");
try (FileWriter fw = new FileWriter(p, true)) {{ try (FileWriter fw = new FileWriter(p, true)) {{
fw.write(line.toString()); fw.write(line.toString());
@ -1015,13 +1022,11 @@ public class NyxHarness {{
public static void main(String[] args) {{ public static void main(String[] args) {{
String payload = System.getenv("NYX_PAYLOAD"); String payload = System.getenv("NYX_PAYLOAD");
if (payload == null) payload = ""; if (payload == null) payload = "";
String rendered = nyxXmlParse(payload); nyxXmlParse(payload);
nyxXxeProbe(rendered, nyxLastExpanded); nyxXxeProbe(payload, nyxLastExpanded);
System.out.println("__NYX_SINK_HIT__"); System.out.println("__NYX_SINK_HIT__");
StringBuilder body = new StringBuilder(64); StringBuilder body = new StringBuilder(64);
body.append("{{\"render\":\""); body.append("{{\"entity_expanded\":").append(nyxLastExpanded ? "true" : "false").append("}}");
nyxJsonEscape(rendered, body);
body.append("\",\"entity_expanded\":").append(nyxLastExpanded ? "true" : "false").append("}}");
System.out.println(body.toString()); System.out.println(body.toString());
}} }}
}} }}

View file

@ -373,11 +373,10 @@ pub fn materialize_node(env: &Environment) -> RuntimeArtifacts {
} }
} }
for fw in &env.frameworks { for fw in &env.frameworks {
if let Some(name) = node_framework_pkg_name(*fw) { if let Some(name) = node_framework_pkg_name(*fw)
if seen.insert(name.to_owned()) { && seen.insert(name.to_owned()) {
deps.push((name.to_owned(), "*")); deps.push((name.to_owned(), "*"));
} }
}
} }
deps.sort_by(|a, b| a.0.cmp(&b.0)); deps.sort_by(|a, b| a.0.cmp(&b.0));

View file

@ -667,14 +667,17 @@ echo json_encode(["render" => $rendered]) . "\n";
} }
} }
/// Phase 05 — Track J.3 XXE harness for PHP (`simplexml_load_string` /// Phase 05 — Track J.3 XXE harness for PHP (`simplexml_load_string`).
/// under `libxml_disable_entity_loader(false)`).
/// ///
/// Reads `NYX_PAYLOAD`, scans for `<!ENTITY name SYSTEM "uri">` /// Reads `NYX_PAYLOAD`, registers a real `libxml_set_external_entity_loader`
/// declarations, expands them inside `&name;` element references /// callback (the canonical PHP hook for external entity resolution),
/// (matching `simplexml_load_string` / `DOMDocument` with the entity /// parses the payload with `simplexml_load_string` under
/// loader re-enabled), and writes a `ProbeKind::Xxe` probe whose /// `LIBXML_NOENT | LIBXML_DTDLOAD` (the configuration real XXE-prone
/// `entity_expanded` flag tracks whether the substitution fired. /// code uses), and writes a `ProbeKind::Xxe` probe whose
/// `entity_expanded` flag tracks whether the loader fired. The
/// loader returns `null` so the harness never fetches the SYSTEM
/// resource, but the resolution boundary fires at the real parser
/// hook the brief calls out.
pub fn emit_xxe_harness(_spec: &HarnessSpec) -> HarnessSource { pub fn emit_xxe_harness(_spec: &HarnessSpec) -> HarnessSource {
let shim = probe_shim(); let shim = probe_shim();
let body = format!( let body = format!(
@ -682,43 +685,47 @@ pub fn emit_xxe_harness(_spec: &HarnessSpec) -> HarnessSource {
// Nyx dynamic harness — XXE simplexml_load_string (Phase 05 / Track J.3). // Nyx dynamic harness — XXE simplexml_load_string (Phase 05 / Track J.3).
{shim} {shim}
function _nyx_libxml_parse(string $payload): array {{ function _nyx_libxml_parse(string $payload): bool {{
$entities = [];
if (preg_match_all('/<!ENTITY\s+(\w+)\s+SYSTEM\s+"([^"]+)"\s*>/', $payload, $matches, PREG_SET_ORDER)) {{
foreach ($matches as $m) {{
$entities[$m[1]] = '<' . $m[2] . '>';
}}
}}
$expanded = false; $expanded = false;
$rendered = preg_replace_callback('/&(\w+);/', function ($m) use ($entities, &$expanded) {{ // Real parser hook: libxml calls this for every <!ENTITY name SYSTEM "uri">
if (array_key_exists($m[1], $entities)) {{ // reference resolved in the document. We mark expanded and
$expanded = true; // return null so the parser does not actually fetch the resource.
return $entities[$m[1]]; libxml_set_external_entity_loader(function ($public, $system, $context) use (&$expanded) {{
}} $expanded = true;
return $m[0]; return null;
}}, $payload) ?? $payload; }});
return [$rendered, $expanded]; $prev_errors = libxml_use_internal_errors(true);
// LIBXML_NOENT enables entity substitution (turning `&xxe;` into
// the resolved body) and LIBXML_DTDLOAD allows the parser to load
// the DTD declarations — the combination real XXE-vulnerable PHP
// code passes to `simplexml_load_string`.
@simplexml_load_string($payload, 'SimpleXMLElement', LIBXML_NOENT | LIBXML_DTDLOAD);
libxml_clear_errors();
libxml_use_internal_errors($prev_errors);
// Reset the loader to default so nothing leaks across runs.
libxml_set_external_entity_loader(null);
return $expanded;
}} }}
function _nyx_xxe_probe(string $rendered, bool $expanded): void {{ function _nyx_xxe_probe(string $payload, bool $expanded): void {{
$p = getenv('NYX_PROBE_PATH'); $p = getenv('NYX_PROBE_PATH');
if ($p === false || $p === '') return; if ($p === false || $p === '') return;
$rec = [ $rec = [
'sink_callee' => 'simplexml_load_string', 'sink_callee' => 'simplexml_load_string',
'args' => [['kind' => 'String', 'value' => $rendered]], 'args' => [['kind' => 'String', 'value' => $payload]],
'captured_at_ns' => (int) hrtime(true), 'captured_at_ns' => (int) hrtime(true),
'payload_id' => (string) (getenv('NYX_PAYLOAD_ID') ?: ''), 'payload_id' => (string) (getenv('NYX_PAYLOAD_ID') ?: ''),
'kind' => ['kind' => 'Xxe', 'entity_expanded' => $expanded], 'kind' => ['kind' => 'Xxe', 'entity_expanded' => $expanded],
'witness' => __nyx_witness('simplexml_load_string', [$rendered]), 'witness' => __nyx_witness('simplexml_load_string', [$payload]),
]; ];
@file_put_contents($p, json_encode($rec) . "\n", FILE_APPEND); @file_put_contents($p, json_encode($rec) . "\n", FILE_APPEND);
}} }}
$payload = (string) (getenv('NYX_PAYLOAD') ?: ''); $payload = (string) (getenv('NYX_PAYLOAD') ?: '');
[$rendered, $expanded] = _nyx_libxml_parse($payload); $expanded = _nyx_libxml_parse($payload);
_nyx_xxe_probe($rendered, $expanded); _nyx_xxe_probe($payload, $expanded);
echo "__NYX_SINK_HIT__\n"; echo "__NYX_SINK_HIT__\n";
echo json_encode(["render" => $rendered, "entity_expanded" => $expanded]) . "\n"; echo json_encode(["entity_expanded" => $expanded]) . "\n";
"# "#
); );
HarnessSource { HarnessSource {

View file

@ -1438,65 +1438,76 @@ if __name__ == "__main__":
/// Phase 05 — Track J.3 XXE harness for Python (`lxml.etree`). /// Phase 05 — Track J.3 XXE harness for Python (`lxml.etree`).
/// ///
/// Reads `NYX_PAYLOAD`, runs a regex-based DOCTYPE/ENTITY scanner that /// Reads `NYX_PAYLOAD`, parses it with `xml.parsers.expat` (the stdlib
/// substitutes any `<!ENTITY name SYSTEM "uri">` body inside `&name;` /// XML parser backing `xml.etree.ElementTree` and `lxml`), installs a
/// element references (matching `lxml.etree.XMLParser(resolve_entities= /// real `ExternalEntityRefHandler` to detect external-entity resolution
/// True)` semantics) and writes a `ProbeKind::Xxe` probe whose /// at the parser hook, and writes a `ProbeKind::Xxe` probe whose
/// `entity_expanded` flag tracks whether the substitution actually /// `entity_expanded` flag tracks whether the handler actually fired.
/// fired. The synthetic resolver keeps the corpus deterministic /// The handler returns an empty replacement so the harness never
/// without bundling lxml in the sandbox image; the harness still /// fetches the SYSTEM resource (sandbox safety) but the resolution
/// exercises the probe-channel, oracle, and differential plumbing /// boundary is exercised at the parser level.
/// end-to-end.
pub fn emit_xxe_harness(_spec: &HarnessSpec) -> HarnessSource { pub fn emit_xxe_harness(_spec: &HarnessSpec) -> HarnessSource {
let probe = probe_shim(); let probe = probe_shim();
let body = format!( let body = format!(
r#"#!/usr/bin/env python3 r#"#!/usr/bin/env python3
"""Nyx dynamic harness — XXE lxml (Phase 05 / Track J.3).""" """Nyx dynamic harness — XXE xml.parsers.expat (Phase 05 / Track J.3)."""
import os, json, re, sys, time import os, json, sys, time
import xml.parsers.expat as _nyx_expat
{probe} {probe}
_NYX_DOCTYPE_ENTITY = re.compile( def _nyx_xxe_parse(payload):
r'<!ENTITY\s+(\w+)\s+SYSTEM\s+"([^"]+)"\s*>' expanded = [False]
) parser = _nyx_expat.ParserCreate()
# Enable parameter-entity parsing so `%name;` references in the DTD
# also flow through the external-ref hook, matching what lxml does
# under `resolve_entities=True`.
try:
parser.SetParamEntityParsing(_nyx_expat.XML_PARAM_ENTITY_PARSING_ALWAYS)
except Exception:
pass
def _nyx_lxml_parse(payload): def _external_ref(context, base, system_id, public_id):
# Parse the payload with `resolve_entities=True` semantics: bind # Real parser hook: fired by expat for every `<!ENTITY x SYSTEM "...">`
# `<!ENTITY name SYSTEM "uri">` declarations into a map then # reference inside element bodies / DTD. Mark expanded and return an
# substitute `&name;` references inside element bodies. # empty replacement so we never actually fetch the SYSTEM resource.
entities = {{}} expanded[0] = True
for m in _NYX_DOCTYPE_ENTITY.finditer(payload): sub = parser.ExternalEntityParserCreate(context, "utf-8")
entities[m.group(1)] = '<' + m.group(2) + '>' try:
expanded = False sub.Parse("", 1)
def _sub(match): except _nyx_expat.ExpatError:
nonlocal expanded pass
name = match.group(1) return 1
if name in entities:
expanded = True
return entities[name]
return match.group(0)
rendered = re.sub(r'&(\w+);', _sub, payload)
return rendered, expanded
def _nyx_xxe_probe(rendered, expanded): parser.ExternalEntityRefHandler = _external_ref
payload_bytes = payload.encode("utf-8", "replace") if isinstance(payload, str) else payload
try:
parser.Parse(payload_bytes, 1)
except _nyx_expat.ExpatError:
# Malformed XML still counts as a parser invocation; expanded
# flag reflects whatever the hook saw before the error.
pass
return expanded[0]
def _nyx_xxe_probe(payload, expanded):
rec = {{ rec = {{
"sink_callee": "lxml.etree.XMLParser.parse", "sink_callee": "lxml.etree.XMLParser.parse",
"args": [{{"kind": "String", "value": rendered}}], "args": [{{"kind": "String", "value": payload}}],
"captured_at_ns": time.time_ns(), "captured_at_ns": time.time_ns(),
"payload_id": os.environ.get("NYX_PAYLOAD_ID", ""), "payload_id": os.environ.get("NYX_PAYLOAD_ID", ""),
"kind": {{"kind": "Xxe", "entity_expanded": bool(expanded)}}, "kind": {{"kind": "Xxe", "entity_expanded": bool(expanded)}},
"witness": __nyx_witness("lxml.etree.XMLParser.parse", [rendered]), "witness": __nyx_witness("lxml.etree.XMLParser.parse", [payload]),
}} }}
__nyx_emit(rec) __nyx_emit(rec)
def _nyx_run(): def _nyx_run():
payload = os.environ.get("NYX_PAYLOAD", "") payload = os.environ.get("NYX_PAYLOAD", "")
rendered, expanded = _nyx_lxml_parse(payload) expanded = _nyx_xxe_parse(payload)
_nyx_xxe_probe(rendered, expanded) _nyx_xxe_probe(payload, expanded)
# Sink-hit sentinel flips SandboxOutcome.sink_hit so the runner's # Sink-hit sentinel flips SandboxOutcome.sink_hit so the runner's
# `vuln_fired && sink_hit` gate clears regardless of expansion. # `vuln_fired && sink_hit` gate clears regardless of expansion.
print("__NYX_SINK_HIT__", flush=True) print("__NYX_SINK_HIT__", flush=True)
sys.stdout.write(json.dumps({{"render": rendered, "entity_expanded": expanded}}) + "\n") sys.stdout.write(json.dumps({{"entity_expanded": bool(expanded)}}) + "\n")
sys.stdout.flush() sys.stdout.flush()
if __name__ == "__main__": if __name__ == "__main__":

View file

@ -972,57 +972,75 @@ STDOUT.flush
/// Phase 05 — Track J.3 XXE harness for Ruby (REXML / Nokogiri). /// Phase 05 — Track J.3 XXE harness for Ruby (REXML / Nokogiri).
/// ///
/// Reads `NYX_PAYLOAD`, scans for `<!ENTITY name SYSTEM "uri">` /// Reads `NYX_PAYLOAD`, parses it with stdlib `REXML::Document.new`,
/// declarations, substitutes them inside `&name;` element bodies, and /// inspects the resulting `doctype.entities` table for SYSTEM/PUBLIC
/// writes a `ProbeKind::Xxe` probe whose `entity_expanded` flag tracks /// external-entity declarations the parser actually parsed and
/// whether the substitution fired. Brief lists a framework adapter /// registered, and writes a `ProbeKind::Xxe` probe whose
/// for Ruby XXE (`xxe_ruby`); the harness keeps the corpus /// `entity_expanded` flag tracks whether REXML registered any
/// end-to-end-exercisable without bundling REXML / Nokogiri. /// external entity. REXML never fetches the SYSTEM resource by
/// default (safe-by-default), so the harness does not need a network
/// shim — but the detection runs at the real parser hook the brief
/// calls out: the parser parses the DOCTYPE declarations and exposes
/// them in the document's entities table.
pub fn emit_xxe_harness(_spec: &HarnessSpec) -> HarnessSource { pub fn emit_xxe_harness(_spec: &HarnessSpec) -> HarnessSource {
let shim = probe_shim(); let shim = probe_shim();
let body = format!( let body = format!(
r#"# Nyx dynamic harness — XXE REXML / Nokogiri (Phase 05 / Track J.3). r#"# Nyx dynamic harness — XXE REXML (Phase 05 / Track J.3).
require 'json' require 'json'
require 'rexml/document'
require 'stringio'
{shim} {shim}
def _nyx_libxml_parse(payload) def _nyx_libxml_parse(payload)
entities = {{}} # Real parser hook: REXML parses `<!ENTITY name SYSTEM "uri">` declarations
payload.scan(/<!ENTITY\s+(\w+)\s+SYSTEM\s+"([^"]+)"\s*>/) do |name, uri| # into Entity objects on the doctype. Inspect the entities table to
entities[name] = "<#{{uri}}>" # detect every external-entity reference the parser registered.
end
expanded = false expanded = false
rendered = payload.gsub(/&(\w+);/) do begin
name = Regexp.last_match(1) doc = REXML::Document.new(payload)
if entities.key?(name) if doc.doctype
expanded = true doc.doctype.entities.each_value do |ent|
entities[name] s = ent.to_s
else if s =~ /SYSTEM|PUBLIC/
Regexp.last_match(0) expanded = true
end
end
# REXML serialization raises on unresolved external entity refs
# in element bodies catch the raise as a secondary signal that
# the parser saw an external reference past the declaration.
begin
doc.write(StringIO.new)
rescue StandardError
expanded = true
end
end end
rescue StandardError
# Malformed XML still counts as a parser invocation; expanded
# reflects whatever the parser saw before the error.
end end
[rendered, expanded] expanded
end end
def _nyx_xxe_probe(rendered, expanded) def _nyx_xxe_probe(payload, expanded)
p = ENV['NYX_PROBE_PATH'] p = ENV['NYX_PROBE_PATH']
return if p.nil? || p.empty? return if p.nil? || p.empty?
rec = {{ rec = {{
'sink_callee' => 'REXML::Document.new', 'sink_callee' => 'REXML::Document.new',
'args' => [{{ 'kind' => 'String', 'value' => rendered }}], 'args' => [{{ 'kind' => 'String', 'value' => payload }}],
'captured_at_ns' => Process.clock_gettime(Process::CLOCK_MONOTONIC, :nanosecond), 'captured_at_ns' => Process.clock_gettime(Process::CLOCK_MONOTONIC, :nanosecond),
'payload_id' => ENV['NYX_PAYLOAD_ID'] || '', 'payload_id' => ENV['NYX_PAYLOAD_ID'] || '',
'kind' => {{ 'kind' => 'Xxe', 'entity_expanded' => !!expanded }}, 'kind' => {{ 'kind' => 'Xxe', 'entity_expanded' => !!expanded }},
'witness' => __nyx_witness('REXML::Document.new', [rendered]), 'witness' => __nyx_witness('REXML::Document.new', [payload]),
}} }}
File.open(p, 'a') {{ |f| f.write(rec.to_json + "\n") }} File.open(p, 'a') {{ |f| f.write(rec.to_json + "\n") }}
end end
payload = ENV['NYX_PAYLOAD'] || '' payload = ENV['NYX_PAYLOAD'] || ''
rendered, expanded = _nyx_libxml_parse(payload) expanded = _nyx_libxml_parse(payload)
_nyx_xxe_probe(rendered, expanded) _nyx_xxe_probe(payload, expanded)
STDOUT.puts '__NYX_SINK_HIT__' STDOUT.puts '__NYX_SINK_HIT__'
STDOUT.puts JSON.generate({{"render" => rendered, "entity_expanded" => expanded}}) STDOUT.puts JSON.generate({{"entity_expanded" => expanded}})
STDOUT.flush STDOUT.flush
"# "#
); );

View file

@ -1078,8 +1078,8 @@ fn class_derives_default(entry_src: &str, class: &str) -> bool {
if boundary_ok { if boundary_ok {
let window_start = decl_pos.saturating_sub(256); let window_start = decl_pos.saturating_sub(256);
let window = &entry_src[window_start..decl_pos]; let window = &entry_src[window_start..decl_pos];
if let Some(derive_pos) = window.rfind("#[derive(") { if let Some(derive_pos) = window.rfind("#[derive(")
if let Some(end_rel) = window[derive_pos..].find(")]") { && let Some(end_rel) = window[derive_pos..].find(")]") {
let end = derive_pos + end_rel; let end = derive_pos + end_rel;
let derive_list = &window[derive_pos + "#[derive(".len()..end]; let derive_list = &window[derive_pos + "#[derive(".len()..end];
let between = &window[end + ")]".len()..]; let between = &window[end + ")]".len()..];
@ -1102,7 +1102,6 @@ fn class_derives_default(entry_src: &str, class: &str) -> bool {
return true; return true;
} }
} }
}
} }
search_from = decl_pos + 1; search_from = decl_pos + 1;
} }

View file

@ -142,13 +142,11 @@ fn handle_connection(stream: TcpStream, hits: Arc<Mutex<HashSet<String>>>) {
let _ = stream.set_read_timeout(Some(Duration::from_secs(2))); let _ = stream.set_read_timeout(Some(Duration::from_secs(2)));
let mut reader = BufReader::new(&stream); let mut reader = BufReader::new(&stream);
let mut first_line = String::new(); let mut first_line = String::new();
if reader.read_line(&mut first_line).is_ok() { if reader.read_line(&mut first_line).is_ok()
if let Some(nonce) = parse_nonce_from_request_line(&first_line) { && let Some(nonce) = parse_nonce_from_request_line(&first_line)
if let Ok(mut h) = hits.lock() { && let Ok(mut h) = hits.lock() {
h.insert(nonce); h.insert(nonce);
} }
}
}
// Drain remaining headers so the client doesn't get ECONNRESET. // Drain remaining headers so the client doesn't get ECONNRESET.
loop { loop {
let mut line = String::new(); let mut line = String::new();

View file

@ -747,11 +747,10 @@ fn stdout_template_equals(stdout: &[u8], expected: u64) -> bool {
let Ok(v) = parsed else { continue }; let Ok(v) = parsed else { continue };
let Some(render) = v.get("render") else { continue }; let Some(render) = v.get("render") else { continue };
let Some(s) = render.as_str() else { continue }; let Some(s) = render.as_str() else { continue };
if let Ok(n) = s.trim().parse::<u64>() { if let Ok(n) = s.trim().parse::<u64>()
if n == expected { && n == expected {
return true; return true;
} }
}
} }
false false
} }
@ -931,7 +930,7 @@ fn extract_redirect_host(location: &str) -> Option<String> {
}; };
// Strip path / query / fragment from the host segment. // Strip path / query / fragment from the host segment.
let end = rest let end = rest
.find(|c: char| matches!(c, '/' | '?' | '#')) .find(['/', '?', '#'])
.unwrap_or(rest.len()); .unwrap_or(rest.len());
let authority = &rest[..end]; let authority = &rest[..end];
// Strip userinfo + port. Bracketed IPv6 authorities (`[::1]` or // Strip userinfo + port. Bracketed IPv6 authorities (`[::1]` or

View file

@ -113,9 +113,11 @@ impl ProbeArg {
/// sink no longer satisfies the oracle. /// sink no longer satisfies the oracle.
#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)] #[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
#[serde(tag = "kind")] #[serde(tag = "kind")]
#[derive(Default)]
pub enum ProbeKind { pub enum ProbeKind {
/// Standard sink observation: arguments were captured before the sink /// Standard sink observation: arguments were captured before the sink
/// returned normally (or raised a non-crash exception). /// returned normally (or raised a non-crash exception).
#[default]
Normal, Normal,
/// Sink invocation was interrupted by a fatal signal that the /// Sink invocation was interrupted by a fatal signal that the
/// sink-site handler intercepted. The captured `signal` is the one /// sink-site handler intercepted. The captured `signal` is the one
@ -305,11 +307,6 @@ pub enum ProbeKind {
}, },
} }
impl Default for ProbeKind {
fn default() -> Self {
ProbeKind::Normal
}
}
/// Bounded forensic snapshot captured alongside a [`SinkProbe`] /// Bounded forensic snapshot captured alongside a [`SinkProbe`]
/// (Phase 08 — Track C.5). /// (Phase 08 — Track C.5).

View file

@ -90,6 +90,7 @@ impl std::fmt::Display for ReproError {
/// ///
/// `harness_source` is the generated harness source code. /// `harness_source` is the generated harness source code.
/// `entry_source` is the extracted entry-point source (may be empty). /// `entry_source` is the extracted entry-point source (may be empty).
#[allow(clippy::too_many_arguments)]
pub fn write( pub fn write(
spec: &HarnessSpec, spec: &HarnessSpec,
opts: &SandboxOptions, opts: &SandboxOptions,
@ -635,7 +636,7 @@ fn repro_readme(spec: &HarnessSpec, verdict: &VerifyResult) -> String {
The expected outcome is in `expected/outcome.json`.\n", The expected outcome is in `expected/outcome.json`.\n",
finding_id = spec.finding_id, finding_id = spec.finding_id,
status = verdict.status, status = verdict.status,
cap = format!("{:?}", spec.expected_cap), cap = format_args!("{:?}", spec.expected_cap),
entry = spec.entry_name, entry = spec.entry_name,
) )
} }

View file

@ -197,14 +197,13 @@ pub fn run_spec(spec: &HarnessSpec, opts: &SandboxOptions) -> Result<RunOutcome,
// non-fatal failures (Io / Unsupported), falling back to system python3. // non-fatal failures (Io / Unsupported), falling back to system python3.
match build_sandbox::prepare_python(spec, &harness.workdir) { match build_sandbox::prepare_python(spec, &harness.workdir) {
Ok(build_result) => { Ok(build_result) => {
if let Some(cmd0) = harness.command.first_mut() { if let Some(cmd0) = harness.command.first_mut()
if cmd0 == "python3" || cmd0 == "python" { && (cmd0 == "python3" || cmd0 == "python") {
let venv_python = build_result.venv_path.join("bin").join("python3"); let venv_python = build_result.venv_path.join("bin").join("python3");
if venv_python.exists() { if venv_python.exists() {
*cmd0 = venv_python.to_string_lossy().into_owned(); *cmd0 = venv_python.to_string_lossy().into_owned();
} }
} }
}
} }
Err(build_sandbox::BuildError::BuildFailed { stderr, attempts }) => { Err(build_sandbox::BuildError::BuildFailed { stderr, attempts }) => {
return Err(RunError::BuildFailed { stderr, attempts }); return Err(RunError::BuildFailed { stderr, attempts });
@ -241,11 +240,8 @@ pub fn run_spec(spec: &HarnessSpec, opts: &SandboxOptions) -> Result<RunOutcome,
} }
Lang::JavaScript | Lang::TypeScript => { Lang::JavaScript | Lang::TypeScript => {
// npm install for dependency resolution (no deps in basic fixtures). // npm install for dependency resolution (no deps in basic fixtures).
match build_sandbox::prepare_node(spec, &harness.workdir) { if let Err(build_sandbox::BuildError::BuildFailed { stderr, attempts }) = build_sandbox::prepare_node(spec, &harness.workdir) {
Err(build_sandbox::BuildError::BuildFailed { stderr, attempts }) => { return Err(RunError::BuildFailed { stderr, attempts });
return Err(RunError::BuildFailed { stderr, attempts });
}
_ => {}
} }
} }
Lang::Go => { Lang::Go => {
@ -288,11 +284,8 @@ pub fn run_spec(spec: &HarnessSpec, opts: &SandboxOptions) -> Result<RunOutcome,
} }
Lang::Php => { Lang::Php => {
// composer install if composer.json is present. // composer install if composer.json is present.
match build_sandbox::prepare_php(spec, &harness.workdir) { if let Err(build_sandbox::BuildError::BuildFailed { stderr, attempts }) = build_sandbox::prepare_php(spec, &harness.workdir) {
Err(build_sandbox::BuildError::BuildFailed { stderr, attempts }) => { return Err(RunError::BuildFailed { stderr, attempts });
return Err(RunError::BuildFailed { stderr, attempts });
}
_ => {}
} }
} }
Lang::C => { Lang::C => {
@ -358,11 +351,10 @@ pub fn run_spec(spec: &HarnessSpec, opts: &SandboxOptions) -> Result<RunOutcome,
// create the file is non-fatal: the legacy `Oracle::OutputContains` // create the file is non-fatal: the legacy `Oracle::OutputContains`
// oracle still works without a channel. // oracle still works without a channel.
let mut effective_opts = opts.clone(); let mut effective_opts = opts.clone();
if effective_opts.probe_channel.is_none() { if effective_opts.probe_channel.is_none()
if let Ok(ch) = ProbeChannel::for_workdir(&harness.workdir) { && let Ok(ch) = ProbeChannel::for_workdir(&harness.workdir) {
effective_opts.probe_channel = Some(Arc::new(ch)); effective_opts.probe_channel = Some(Arc::new(ch));
} }
}
let probe_channel: Option<Arc<ProbeChannel>> = effective_opts.probe_channel.clone(); let probe_channel: Option<Arc<ProbeChannel>> = effective_opts.probe_channel.clone();
// Run only vuln (non-benign) payloads in the main loop. // Run only vuln (non-benign) payloads in the main loop.

View file

@ -277,16 +277,13 @@ pub struct SandboxOptions {
/// Each primitive is best-effort; failures degrade to /// Each primitive is best-effort; failures degrade to
/// [`HardeningLevel::Partial`] without aborting the run. /// [`HardeningLevel::Partial`] without aborting the run.
#[derive(Debug, Clone, Copy, PartialEq, Eq)] #[derive(Debug, Clone, Copy, PartialEq, Eq)]
#[derive(Default)]
pub enum ProcessHardeningProfile { pub enum ProcessHardeningProfile {
#[default]
Standard, Standard,
Strict, Strict,
} }
impl Default for ProcessHardeningProfile {
fn default() -> Self {
ProcessHardeningProfile::Standard
}
}
/// Phase 20 follow-up (Track E.4 ablation harness): selectively skip or /// Phase 20 follow-up (Track E.4 ablation harness): selectively skip or
/// loosen individual Strict-profile primitives so the escape-fixture /// loosen individual Strict-profile primitives so the escape-fixture
@ -419,7 +416,9 @@ impl HostPort {
/// with no egress filter. Reserved for diagnostic / dev-only runs; /// with no egress filter. Reserved for diagnostic / dev-only runs;
/// the verifier never sets this in production. /// the verifier never sets this in production.
#[derive(Debug, Clone)] #[derive(Debug, Clone)]
#[derive(Default)]
pub enum NetworkPolicy { pub enum NetworkPolicy {
#[default]
None, None,
StubsOnly { allow: Vec<HostPort> }, StubsOnly { allow: Vec<HostPort> },
OobOutbound { listener: Arc<OobListener> }, OobOutbound { listener: Arc<OobListener> },
@ -461,11 +460,6 @@ impl NetworkPolicy {
} }
} }
impl Default for NetworkPolicy {
fn default() -> Self {
NetworkPolicy::None
}
}
#[derive(Debug, Clone, Copy, PartialEq, Eq)] #[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum SandboxBackend { pub enum SandboxBackend {
@ -882,8 +876,8 @@ fn rewrite_extra_env_for_container(
extra_env extra_env
.iter() .iter()
.map(|(k, v)| { .map(|(k, v)| {
if k == "NYX_FS_ROOT" { if k == "NYX_FS_ROOT"
if let Some(idx) = fs_stub_roots && let Some(idx) = fs_stub_roots
.iter() .iter()
.position(|p| p.as_os_str() == std::ffi::OsStr::new(v)) .position(|p| p.as_os_str() == std::ffi::OsStr::new(v))
{ {
@ -892,7 +886,6 @@ fn rewrite_extra_env_for_container(
format!("{}/{idx}", docker::STUB_MOUNT_ROOT), format!("{}/{idx}", docker::STUB_MOUNT_ROOT),
); );
} }
}
(k.clone(), v.clone()) (k.clone(), v.clone())
}) })
.collect() .collect()
@ -1163,12 +1156,11 @@ fn exec_in_container(
// fixture the process backend confirms. Falls through silently for // fixture the process backend confirms. Falls through silently for
// non-UTF-8 payloads (a `docker -e` argument must be valid UTF-8), // non-UTF-8 payloads (a `docker -e` argument must be valid UTF-8),
// leaving consumers to decode `NYX_PAYLOAD_B64` themselves. // leaving consumers to decode `NYX_PAYLOAD_B64` themselves.
if let Ok(s) = std::str::from_utf8(payload_bytes) { if let Ok(s) = std::str::from_utf8(payload_bytes)
if !s.contains('\0') { && !s.contains('\0') {
cmd_args.push("-e".into()); cmd_args.push("-e".into());
cmd_args.push(format!("NYX_PAYLOAD={s}")); cmd_args.push(format!("NYX_PAYLOAD={s}"));
} }
}
// Forward harness-specific env vars. // Forward harness-specific env vars.
for (k, v) in &harness.env { for (k, v) in &harness.env {
cmd_args.push("-e".into()); cmd_args.push("-e".into());
@ -1750,7 +1742,7 @@ fn contains_subslice(hay: &[u8], needle: &[u8]) -> bool {
fn base64_encode(data: &[u8]) -> String { fn base64_encode(data: &[u8]) -> String {
const ALPHABET: &[u8] = b"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; const ALPHABET: &[u8] = b"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
let mut out = String::with_capacity((data.len() + 2) / 3 * 4); let mut out = String::with_capacity(data.len().div_ceil(3) * 4);
for chunk in data.chunks(3) { for chunk in data.chunks(3) {
let b0 = chunk[0] as u32; let b0 = chunk[0] as u32;
let b1 = if chunk.len() > 1 { chunk[1] as u32 } else { 0 }; let b1 = if chunk.len() > 1 { chunk[1] as u32 } else { 0 };

View file

@ -312,11 +312,10 @@ impl HarnessSpec {
// priority — calling them here would short-circuit the more precise // priority — calling them here would short-circuit the more precise
// strategies (FromFlowSteps / FromRuleNamespace / FromFuncSummaryAuto) // strategies (FromFlowSteps / FromRuleNamespace / FromFuncSummaryAuto)
// whenever the rule id happens to contain `.http.` / `.cli.`. // whenever the rule id happens to contain `.http.` / `.cli.`.
if let (Some(s), Some(cg)) = (summaries, callgraph) { if let (Some(s), Some(cg)) = (summaries, callgraph)
if let Some(spec) = derive_from_callgraph_walk_only(diag, evidence, s, cg) { && let Some(spec) = derive_from_callgraph_walk_only(diag, evidence, s, cg) {
return Ok(spec); return Ok(spec);
} }
}
// Try each strategy in priority order; first non-None wins. // Try each strategy in priority order; first non-None wins.
if let Some(spec) = derive_from_flow_steps(diag, evidence, summaries) { if let Some(spec) = derive_from_flow_steps(diag, evidence, summaries) {
@ -520,11 +519,10 @@ pub fn derive_from_rule_namespace_with(
// Cross-check: the diag's file extension must agree with the rule's // Cross-check: the diag's file extension must agree with the rule's
// language prefix when both are available. Disagreement is a stronger // language prefix when both are available. Disagreement is a stronger
// signal of a mis-rooted finding than a missing extension. // signal of a mis-rooted finding than a missing extension.
if let Some(path_lang) = lang_from_path(&diag.path) { if let Some(path_lang) = lang_from_path(&diag.path)
if path_lang != lang { && path_lang != lang {
return None; return None;
} }
}
let entry_function = resolve_enclosing_function(diag, evidence, summaries, lang) let entry_function = resolve_enclosing_function(diag, evidence, summaries, lang)
.unwrap_or_else(|| "<unknown>".to_owned()); .unwrap_or_else(|| "<unknown>".to_owned());
@ -750,8 +748,8 @@ pub fn derive_from_callgraph_entry_full(
// Step 0: callgraph-aware reverse-edge walk to the nearest entry-point // Step 0: callgraph-aware reverse-edge walk to the nearest entry-point
// ancestor. Only fires when both summaries *and* callgraph are present. // ancestor. Only fires when both summaries *and* callgraph are present.
if let (Some(s), Some(cg)) = (summaries, callgraph) { if let (Some(s), Some(cg)) = (summaries, callgraph)
if let Some(found) = find_entry_via_callgraph(diag, evidence, s, cg, lang) { && let Some(found) = find_entry_via_callgraph(diag, evidence, s, cg, lang) {
let entry_kind = found let entry_kind = found
.summary .summary
.entry_kind .entry_kind
@ -778,7 +776,6 @@ pub fn derive_from_callgraph_entry_full(
spec.spec_hash = compute_spec_hash(&spec); spec.spec_hash = compute_spec_hash(&spec);
return Some(spec); return Some(spec);
} }
}
// Step 1: try summary-based classification of the enclosing function. // Step 1: try summary-based classification of the enclosing function.
let summary_kind = enclosing_function_from_flow_steps(evidence) let summary_kind = enclosing_function_from_flow_steps(evidence)
@ -936,14 +933,13 @@ fn find_entry_via_callgraph<'a>(
continue; continue;
} }
let caller_key = &callgraph.graph[caller_node]; let caller_key = &callgraph.graph[caller_node];
if let Some(caller_summary) = summaries.get(caller_key) { if let Some(caller_summary) = summaries.get(caller_key)
if is_entry_point(caller_summary, callgraph) { && is_entry_point(caller_summary, callgraph) {
return Some(EntryHit { return Some(EntryHit {
key: caller_key.clone(), key: caller_key.clone(),
summary: caller_summary, summary: caller_summary,
}); });
} }
}
queue.push_back(caller_node); queue.push_back(caller_node);
} }
} }
@ -973,11 +969,10 @@ fn entry_kind_from_summary(_kind: &crate::entry_points::EntryKind) -> EntryKind
/// resolve when the extension is well-known. /// resolve when the extension is well-known.
fn lang_from_path(path: &str) -> Option<Lang> { fn lang_from_path(path: &str) -> Option<Lang> {
let p = Path::new(path); let p = Path::new(path);
if let Some(ext) = p.extension().and_then(|e| e.to_str()) { if let Some(ext) = p.extension().and_then(|e| e.to_str())
if let Some(lang) = Lang::from_extension(ext) { && let Some(lang) = Lang::from_extension(ext) {
return Some(lang); return Some(lang);
} }
}
// Fall back to a shebang / content sniff over the file head. // Fall back to a shebang / content sniff over the file head.
let head = read_file_head(p, 200); let head = read_file_head(p, 200);
if head.is_empty() { if head.is_empty() {
@ -1308,16 +1303,14 @@ fn lang_slug(lang: Lang) -> &'static str {
/// outermost callable that receives the tainted input. /// outermost callable that receives the tainted input.
pub fn outermost_entry(steps: &[crate::evidence::FlowStep]) -> Option<EntryRef> { pub fn outermost_entry(steps: &[crate::evidence::FlowStep]) -> Option<EntryRef> {
for step in steps { for step in steps {
if matches!(step.kind, FlowStepKind::Source) { if matches!(step.kind, FlowStepKind::Source)
if let Some(ref func) = step.function { && let Some(ref func) = step.function
if !func.is_empty() { && !func.is_empty() {
return Some(EntryRef { return Some(EntryRef {
file: step.file.clone(), file: step.file.clone(),
function: func.clone(), function: func.clone(),
}); });
} }
}
}
} }
None None
} }
@ -1340,10 +1333,9 @@ pub fn default_toolchain_id(lang: Lang) -> &'static str {
/// Blake3 hash of the spec's key fields, truncated to 8 bytes and hex-encoded. /// Blake3 hash of the spec's key fields, truncated to 8 bytes and hex-encoded.
/// ///
/// Inputs (in order): /// Inputs (in order): [`SPEC_FORMAT_VERSION`] (u32 LE), entry_file,
/// `SPEC_FORMAT_VERSION` (u32 LE), entry_file, entry_name, payload_slot tag /// entry_name, payload_slot tag + value, expected_cap bits (u32 LE),
/// + value, expected_cap bits (u32 LE), sorted constraint_hints, /// sorted constraint_hints, toolchain_id, [`CORPUS_VERSION`] (u32 LE).
/// toolchain_id, `CORPUS_VERSION` (u32 LE).
/// ///
/// Bump [`SPEC_FORMAT_VERSION`] when the inputs or semantics change. /// Bump [`SPEC_FORMAT_VERSION`] when the inputs or semantics change.
fn compute_spec_hash(spec: &HarnessSpec) -> String { fn compute_spec_hash(spec: &HarnessSpec) -> String {

View file

@ -226,11 +226,10 @@ fn accept_loop(
let _ = stream.set_read_timeout(Some(Duration::from_secs(2))); let _ = stream.set_read_timeout(Some(Duration::from_secs(2)));
let _ = stream.set_write_timeout(Some(Duration::from_secs(2))); let _ = stream.set_write_timeout(Some(Duration::from_secs(2)));
if let Some(ev) = handle_connection(stream, MAX_REQUEST_BYTES) { if let Some(ev) = handle_connection(stream, MAX_REQUEST_BYTES)
if let Ok(mut g) = events.lock() { && let Ok(mut g) = events.lock() {
g.push(ev); g.push(ev);
} }
}
} }
} }
@ -261,21 +260,18 @@ fn handle_connection(mut stream: TcpStream, max_bytes: usize) -> Option<StubEven
if let Some(rest) = trimmed if let Some(rest) = trimmed
.to_ascii_lowercase() .to_ascii_lowercase()
.strip_prefix("content-length:") .strip_prefix("content-length:")
{ && let Ok(n) = rest.trim().parse::<usize>() {
if let Ok(n) = rest.trim().parse::<usize>() {
content_length = n.min(max_bytes); content_length = n.min(max_bytes);
} }
}
headers.push(trimmed.to_owned()); headers.push(trimmed.to_owned());
} }
// Body, capped at content_length (already clamped to max_bytes). // Body, capped at content_length (already clamped to max_bytes).
let mut body = vec![0u8; content_length]; let mut body = vec![0u8; content_length];
if content_length > 0 { if content_length > 0
if reader.read_exact(&mut body).is_err() { && reader.read_exact(&mut body).is_err() {
body.clear(); body.clear();
} }
}
// Always reply 200 OK with no body. // Always reply 200 OK with no body.
let _ = stream.write_all(b"HTTP/1.1 200 OK\r\nContent-Length: 0\r\n\r\n"); let _ = stream.write_all(b"HTTP/1.1 200 OK\r\nContent-Length: 0\r\n\r\n");

View file

@ -115,11 +115,10 @@ fn try_rust_toolchain_toml(root: &Path) -> Option<ToolchainResolution> {
if line.starts_with('[') { if line.starts_with('[') {
in_toolchain = false; in_toolchain = false;
} }
if in_toolchain && line.starts_with("channel") { if in_toolchain && line.starts_with("channel")
if let Some(ver) = extract_version_from_toml_value(line) { && let Some(ver) = extract_version_from_toml_value(line) {
return Some(map_rust_version(&ver, RustPinOrigin::RustToolchainToml)); return Some(map_rust_version(&ver, RustPinOrigin::RustToolchainToml));
} }
}
} }
None None
} }
@ -138,11 +137,10 @@ fn try_cargo_toml_rust_version(root: &Path) -> Option<ToolchainResolution> {
let content = std::fs::read_to_string(root.join("Cargo.toml")).ok()?; let content = std::fs::read_to_string(root.join("Cargo.toml")).ok()?;
for line in content.lines() { for line in content.lines() {
let line = line.trim(); let line = line.trim();
if line.starts_with("rust-version") { if line.starts_with("rust-version")
if let Some(ver) = extract_version_from_toml_value(line) { && let Some(ver) = extract_version_from_toml_value(line) {
return Some(map_rust_version(&ver, RustPinOrigin::CargoToml)); return Some(map_rust_version(&ver, RustPinOrigin::CargoToml));
} }
}
} }
None None
} }
@ -248,11 +246,10 @@ fn try_pyproject_toml(root: &Path) -> Option<ToolchainResolution> {
// Look for `requires-python = ">=3.11"` or `python = "3.11"`. // Look for `requires-python = ">=3.11"` or `python = "3.11"`.
for line in content.lines() { for line in content.lines() {
let line = line.trim(); let line = line.trim();
if line.starts_with("requires-python") || (line.starts_with("python") && line.contains('=') && !line.starts_with("python_requires")) { if (line.starts_with("requires-python") || (line.starts_with("python") && line.contains('=') && !line.starts_with("python_requires")))
if let Some(ver) = extract_version_from_toml_value(line) { && let Some(ver) = extract_version_from_toml_value(line) {
return Some(map_version(&ver, PinOrigin::PyprojectToml)); return Some(map_version(&ver, PinOrigin::PyprojectToml));
} }
}
} }
None None
} }
@ -269,11 +266,10 @@ fn try_pipfile(root: &Path) -> Option<ToolchainResolution> {
if line.starts_with('[') { if line.starts_with('[') {
in_requires = false; in_requires = false;
} }
if in_requires && line.starts_with("python_version") { if in_requires && line.starts_with("python_version")
if let Some(ver) = extract_version_from_toml_value(line) { && let Some(ver) = extract_version_from_toml_value(line) {
return Some(map_version(&ver, PinOrigin::Pipfile)); return Some(map_version(&ver, PinOrigin::Pipfile));
} }
}
} }
None None
} }
@ -302,7 +298,7 @@ fn default_python() -> ToolchainResolution {
/// `requires-python = ">=3.11"` → `"3.11"` /// `requires-python = ">=3.11"` → `"3.11"`
/// `python_version = "3.11"` → `"3.11"` /// `python_version = "3.11"` → `"3.11"`
fn extract_version_from_toml_value(line: &str) -> Option<String> { fn extract_version_from_toml_value(line: &str) -> Option<String> {
let after_eq = line.splitn(2, '=').nth(1)?; let after_eq = line.split_once('=')?.1;
let raw = after_eq.trim().trim_matches('"').trim_matches('\''); let raw = after_eq.trim().trim_matches('"').trim_matches('\'');
if raw.is_empty() { if raw.is_empty() {
return None; return None;
@ -335,7 +331,7 @@ fn map_version(version: &str, origin: PinOrigin) -> ToolchainResolution {
("3", Some("12")) => ("python-3.12".to_owned(), false), ("3", Some("12")) => ("python-3.12".to_owned(), false),
("3", Some("13")) => ("python-3.13".to_owned(), false), ("3", Some("13")) => ("python-3.13".to_owned(), false),
// Older 3.x → nearest supported is 3.8 // Older 3.x → nearest supported is 3.8
("3", Some(m)) if m.parse::<u32>().map_or(false, |v| v < 8) => { ("3", Some(m)) if m.parse::<u32>().is_ok_and(|v| v < 8) => {
("python-3.8".to_owned(), true) ("python-3.8".to_owned(), true)
} }
// Newer 3.x beyond catalog → use 3.13 as closest // Newer 3.x beyond catalog → use 3.13 as closest
@ -466,7 +462,7 @@ fn json_line_has_key(line: &str, key: &str) -> bool {
/// Extract a version string from a JSON value like `">=18"` or `"20.x"`. /// Extract a version string from a JSON value like `">=18"` or `"20.x"`.
fn extract_version_from_json_value(line: &str) -> Option<String> { fn extract_version_from_json_value(line: &str) -> Option<String> {
// Find the second quoted value after the colon. // Find the second quoted value after the colon.
let after_colon = line.splitn(2, ':').nth(1)?; let after_colon = line.split_once(':')?.1;
let raw = after_colon.trim().trim_matches('"').trim_matches('\''); let raw = after_colon.trim().trim_matches('"').trim_matches('\'');
let ver = raw.trim_start_matches(|c: char| !c.is_ascii_digit()); let ver = raw.trim_start_matches(|c: char| !c.is_ascii_digit());
// Strip trailing junk: stop at the first char that isn't a version char. // Strip trailing junk: stop at the first char that isn't a version char.
@ -535,10 +531,10 @@ fn map_go_version(version: &str, origin: PinOrigin) -> ToolchainResolution {
("1", Some("21")) => ("go-1.21".to_owned(), false), ("1", Some("21")) => ("go-1.21".to_owned(), false),
("1", Some("22")) => ("go-1.22".to_owned(), false), ("1", Some("22")) => ("go-1.22".to_owned(), false),
("1", Some("23")) => ("go-1.23".to_owned(), false), ("1", Some("23")) => ("go-1.23".to_owned(), false),
("1", Some(m)) if m.parse::<u32>().map_or(false, |v| v >= 24) => { ("1", Some(m)) if m.parse::<u32>().is_ok_and(|v| v >= 24) => {
(format!("go-1.{m}"), true) (format!("go-1.{m}"), true)
} }
("1", Some(m)) if m.parse::<u32>().map_or(false, |v| v < 21) => { ("1", Some(m)) if m.parse::<u32>().is_ok_and(|v| v < 21) => {
(format!("go-1.{m}"), true) (format!("go-1.{m}"), true)
} }
_ => ("go-stable".to_owned(), false), _ => ("go-stable".to_owned(), false),
@ -575,14 +571,13 @@ fn try_pom_xml(root: &Path) -> Option<ToolchainResolution> {
for line in content.lines() { for line in content.lines() {
let trimmed = line.trim(); let trimmed = line.trim();
for tag in &["<java.version>", "<maven.compiler.source>", "<maven.compiler.release>"] { for tag in &["<java.version>", "<maven.compiler.source>", "<maven.compiler.release>"] {
if trimmed.starts_with(tag) { if trimmed.starts_with(tag)
if let Some(inner) = trimmed.strip_prefix(tag) { && let Some(inner) = trimmed.strip_prefix(tag) {
let version = inner.split('<').next().unwrap_or("").trim(); let version = inner.split('<').next().unwrap_or("").trim();
if !version.is_empty() { if !version.is_empty() {
return Some(map_java_version(version, PinOrigin::PomXml)); return Some(map_java_version(version, PinOrigin::PomXml));
} }
} }
}
} }
} }
None None
@ -597,11 +592,10 @@ fn try_build_gradle(root: &Path) -> Option<ToolchainResolution> {
let trimmed = line.trim(); let trimmed = line.trim();
// Groovy: sourceCompatibility = '21' or JavaVersion.VERSION_21 // Groovy: sourceCompatibility = '21' or JavaVersion.VERSION_21
// Kotlin: sourceCompatibility = JavaVersion.VERSION_21 // Kotlin: sourceCompatibility = JavaVersion.VERSION_21
if trimmed.starts_with("sourceCompatibility") || trimmed.starts_with("languageVersion") { if (trimmed.starts_with("sourceCompatibility") || trimmed.starts_with("languageVersion"))
if let Some(ver) = extract_java_version_from_gradle_line(trimmed) { && let Some(ver) = extract_java_version_from_gradle_line(trimmed) {
return Some(map_java_version(&ver, PinOrigin::BuildGradle)); return Some(map_java_version(&ver, PinOrigin::BuildGradle));
} }
}
} }
} }
None None
@ -610,7 +604,7 @@ fn try_build_gradle(root: &Path) -> Option<ToolchainResolution> {
fn extract_java_version_from_gradle_line(line: &str) -> Option<String> { fn extract_java_version_from_gradle_line(line: &str) -> Option<String> {
// Handle: sourceCompatibility = '21' or sourceCompatibility = 21 // Handle: sourceCompatibility = '21' or sourceCompatibility = 21
// and: languageVersion.set(JavaLanguageVersion.of(21)) // and: languageVersion.set(JavaLanguageVersion.of(21))
let after_eq = line.splitn(2, '=').nth(1).unwrap_or(line); let after_eq = line.split_once('=').map(|x| x.1).unwrap_or(line);
// Try to find a number in the value. // Try to find a number in the value.
let digits: String = after_eq.chars() let digits: String = after_eq.chars()
.skip_while(|c| !c.is_ascii_digit()) .skip_while(|c| !c.is_ascii_digit())
@ -687,13 +681,12 @@ fn try_composer_json(root: &Path) -> Option<ToolchainResolution> {
if json_line_has_key(trimmed, "require") { if json_line_has_key(trimmed, "require") {
in_require = true; in_require = true;
} }
if in_require && trimmed.contains("\"php\"") { if in_require && trimmed.contains("\"php\"")
if let Some(ver) = extract_version_from_json_value(trimmed) { && let Some(ver) = extract_version_from_json_value(trimmed) {
return Some(map_php_version(&ver, PinOrigin::ComposerJson)); return Some(map_php_version(&ver, PinOrigin::ComposerJson));
} }
}
// Stop at closing brace of require block. // Stop at closing brace of require block.
if in_require && trimmed == "}," || (in_require && trimmed == "}") { if in_require && (trimmed == "}," || trimmed == "}") {
in_require = false; in_require = false;
} }
} }

View file

@ -713,8 +713,8 @@ pub fn verify_finding(diag: &Diag, opts: &VerifyOptions) -> VerifyResult {
// Verdict cache lookup (§12 Q5): skip execution when a valid cached result exists. // Verdict cache lookup (§12 Q5): skip execution when a valid cached result exists.
let entry_hash = compute_entry_content_hash(&spec.entry_file); let entry_hash = compute_entry_content_hash(&spec.entry_file);
let import_digest = transitive_import_digest_placeholder(); let import_digest = transitive_import_digest_placeholder();
if let Some(ref db_path) = opts.db_path { if let Some(ref db_path) = opts.db_path
if let Some(cached) = lookup_verdict_cache( && let Some(cached) = lookup_verdict_cache(
db_path, db_path,
&spec.spec_hash, &spec.spec_hash,
&entry_hash, &entry_hash,
@ -723,7 +723,6 @@ pub fn verify_finding(diag: &Diag, opts: &VerifyOptions) -> VerifyResult {
) { ) {
return cached; return cached;
} }
}
// Phase 10 (Track D.3): spawn the boundary stubs the spec // Phase 10 (Track D.3): spawn the boundary stubs the spec
// demands *before* the sandbox runs. When `stubs_required` is // demands *before* the sandbox runs. When `stubs_required` is
@ -998,14 +997,14 @@ fn build_verdict(
); );
// If repro write fails, downgrade to NonReproducible. // If repro write fails, downgrade to NonReproducible.
if repro_result.is_err() { if let Err(err) = repro_result {
return VerifyResult { return VerifyResult {
finding_id: finding_id.to_owned(), finding_id: finding_id.to_owned(),
status: VerifyStatus::Inconclusive, status: VerifyStatus::Inconclusive,
triggered_payload: None, triggered_payload: None,
reason: None, reason: None,
inconclusive_reason: Some(InconclusiveReason::NonReproducible), inconclusive_reason: Some(InconclusiveReason::NonReproducible),
detail: Some(format!("repro write failed: {}", repro_result.unwrap_err())), detail: Some(format!("repro write failed: {err}")),
attempts, attempts,
toolchain_match: Some(toolchain_match.to_owned()), toolchain_match: Some(toolchain_match.to_owned()),
differential: run.differential, differential: run.differential,

View file

@ -315,11 +315,10 @@ pub fn build_sarif_with_chains(
// this finding participates in (if any). Stable across // this finding participates in (if any). Stable across
// reruns because both the finding's `stable_hash` and the // reruns because both the finding's `stable_hash` and the
// chain's `stable_hash` are byte-deterministic. // chain's `stable_hash` are byte-deterministic.
if d.stable_hash != 0 { if d.stable_hash != 0
if let Some(chain_hash) = chain_member_of.get(&d.stable_hash) { && let Some(chain_hash) = chain_member_of.get(&d.stable_hash) {
props.insert("chain_member_of".into(), json!(chain_hash)); props.insert("chain_member_of".into(), json!(chain_hash));
} }
}
result["properties"] = Value::Object(props); result["properties"] = Value::Object(props);

View file

@ -40,8 +40,8 @@ pub fn detect_rails_routes(
fn detect_routes_dsl(root: Node, bytes: &[u8], file_rel: &str, out: &mut Vec<SurfaceNode>) { fn detect_routes_dsl(root: Node, bytes: &[u8], file_rel: &str, out: &mut Vec<SurfaceNode>) {
fn recurse(node: Node, bytes: &[u8], file_rel: &str, out: &mut Vec<SurfaceNode>) { fn recurse(node: Node, bytes: &[u8], file_rel: &str, out: &mut Vec<SurfaceNode>) {
if matches!(node.kind(), "call" | "method_call") { if matches!(node.kind(), "call" | "method_call")
if let Some(method_node) = node.child_by_field_name("method") && let Some(method_node) = node.child_by_field_name("method")
&& let Ok(method_text) = method_node.utf8_text(bytes) && let Ok(method_text) = method_node.utf8_text(bytes)
&& let Some((_, method)) = VERBS.iter().find(|(v, _)| *v == method_text) && let Some((_, method)) = VERBS.iter().find(|(v, _)| *v == method_text)
{ {
@ -73,7 +73,6 @@ fn detect_routes_dsl(root: Node, bytes: &[u8], file_rel: &str, out: &mut Vec<Sur
} }
} }
} }
}
let mut cursor = node.walk(); let mut cursor = node.walk();
for child in node.children(&mut cursor) { for child in node.children(&mut cursor) {
recurse(child, bytes, file_rel, out); recurse(child, bytes, file_rel, out);

View file

@ -53,9 +53,9 @@ fn is_app_router_route(path: &Path) -> bool {
} }
fn is_pages_api_route(path: &Path) -> bool { fn is_pages_api_route(path: &Path) -> bool {
let mut comps = path.components().peekable(); let comps = path.components().peekable();
let mut saw_pages = false; let mut saw_pages = false;
while let Some(c) = comps.next() { for c in comps {
if c.as_os_str().to_string_lossy() == "pages" { if c.as_os_str().to_string_lossy() == "pages" {
saw_pages = true; saw_pages = true;
} else if saw_pages && c.as_os_str().to_string_lossy() == "api" { } else if saw_pages && c.as_os_str().to_string_lossy() == "api" {

View file

@ -341,11 +341,10 @@ impl SurfaceMap {
/// Returns the absolute path verbatim when the file is outside the /// Returns the absolute path verbatim when the file is outside the
/// scan root or when path stripping fails. /// scan root or when path stripping fails.
pub fn relative_path_string(path: &Path, scan_root: Option<&Path>) -> String { pub fn relative_path_string(path: &Path, scan_root: Option<&Path>) -> String {
if let Some(root) = scan_root { if let Some(root) = scan_root
if let Ok(rel) = path.strip_prefix(root) { && let Ok(rel) = path.strip_prefix(root) {
return rel.to_string_lossy().replace('\\', "/"); return rel.to_string_lossy().replace('\\', "/");
} }
}
path.to_string_lossy().replace('\\', "/") path.to_string_lossy().replace('\\', "/")
} }

View file

@ -114,11 +114,10 @@ impl Lang {
/// Used by [`crate::dynamic::spec`] so spec derivation no longer rejects /// Used by [`crate::dynamic::spec`] so spec derivation no longer rejects
/// CLI entry points and other extensionless / non-canonical files. /// CLI entry points and other extensionless / non-canonical files.
pub fn from_path_or_content(path: &Path, head_bytes: &[u8]) -> Option<Lang> { pub fn from_path_or_content(path: &Path, head_bytes: &[u8]) -> Option<Lang> {
if let Some(ext) = path.extension().and_then(|e| e.to_str()) { if let Some(ext) = path.extension().and_then(|e| e.to_str())
if let Some(lang) = Self::from_extension(ext) { && let Some(lang) = Self::from_extension(ext) {
return Some(lang); return Some(lang);
} }
}
if let Some(lang) = lang_from_shebang(head_bytes) { if let Some(lang) = lang_from_shebang(head_bytes) {
return Some(lang); return Some(lang);
} }

View file

@ -256,7 +256,7 @@ pub struct FixtureSpec<'a> {
/// ///
/// Captures the fields a regression test must pin: status + typed reasons /// Captures the fields a regression test must pin: status + typed reasons
/// + whether a payload triggered. Excludes machine-dependent fields /// + whether a payload triggered. Excludes machine-dependent fields
/// (`finding_id`, `detail`, `attempts`, `toolchain_match`). /// (`finding_id`, `detail`, `attempts`, `toolchain_match`).
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)] #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct GoldenVerdict { pub struct GoldenVerdict {
pub status: VerifyStatus, pub status: VerifyStatus,

View file

@ -24,11 +24,13 @@ use std::collections::BTreeSet;
const RUN_COUNT: usize = 10; const RUN_COUNT: usize = 10;
fn deny_diag(stable_hash: u64) -> Diag { fn deny_diag(stable_hash: u64) -> Diag {
let mut ev = Evidence::default();
// Triggers the credentials deny rule via the AWS-key regex from // Triggers the credentials deny rule via the AWS-key regex from
// `crate::utils::redact::contains_secret`. The deny rule fires // `crate::utils::redact::contains_secret`. The deny rule fires
// deterministically because the rule lookup table is `const`. // deterministically because the rule lookup table is `const`.
ev.notes = vec!["secret=AKIAFAKEDETERM00000000".to_owned()]; let ev = Evidence {
notes: vec!["secret=AKIAFAKEDETERM00000000".to_owned()],
..Evidence::default()
};
Diag { Diag {
path: "src/handler.py".to_owned(), path: "src/handler.py".to_owned(),
line: 42, line: 42,
@ -84,9 +86,11 @@ fn ten_runs_produce_byte_identical_telemetry_minus_timestamps() {
let diag = deny_diag(0x0123_4567_89ab_cdef); let diag = deny_diag(0x0123_4567_89ab_cdef);
let mut opts = VerifyOptions::default(); let opts = VerifyOptions {
opts.telemetry_policy = SamplingPolicy::keep_all(); telemetry_policy: SamplingPolicy::keep_all(),
opts.trace_verbose = false; trace_verbose: false,
..VerifyOptions::default()
};
let mut verdict_jsons: BTreeSet<String> = BTreeSet::new(); let mut verdict_jsons: BTreeSet<String> = BTreeSet::new();
for _ in 0..RUN_COUNT { for _ in 0..RUN_COUNT {

View file

@ -127,11 +127,10 @@ mod parity_tests {
// BackendUnavailable into Unsupported OR Inconclusive depending on // BackendUnavailable into Unsupported OR Inconclusive depending on
// where the error surfaces, so the skip predicate looks at the // where the error surfaces, so the skip predicate looks at the
// reason text, not the verdict status. // reason text, not the verdict status.
if let Some(ref r) = docker_result.reason { if let Some(ref r) = docker_result.reason
if format!("{r:?}").contains("BackendUnavailable") { && format!("{r:?}").contains("BackendUnavailable") {
return; // Docker absent — skip comparison. return; // Docker absent — skip comparison.
} }
}
assert_eq!( assert_eq!(
process_result.status, docker_result.status, process_result.status, docker_result.status,

View file

@ -189,8 +189,10 @@ mod verify_e2e {
let diag = taint_diag_with_cap(Cap::CRYPTO); let diag = taint_diag_with_cap(Cap::CRYPTO);
let trace = Arc::new(VerifyTrace::new()); let trace = Arc::new(VerifyTrace::new());
let mut opts = VerifyOptions::default(); let opts = VerifyOptions {
opts.trace_sink = Some(Arc::clone(&trace)); trace_sink: Some(Arc::clone(&trace)),
..VerifyOptions::default()
};
let _result = verify_finding(&diag, &opts); let _result = verify_finding(&diag, &opts);

View file

@ -33,7 +33,7 @@ fn scan_with_hashes(dir: &Path) -> Vec<nyx_scanner::commands::scan::Diag> {
/// Attach a simulated dynamic verdict to every finding in the list. /// Attach a simulated dynamic verdict to every finding in the list.
fn set_verdict( fn set_verdict(
diags: &mut Vec<nyx_scanner::commands::scan::Diag>, diags: &mut [nyx_scanner::commands::scan::Diag],
status: VerifyStatus, status: VerifyStatus,
) { ) {
for d in diags.iter_mut() { for d in diags.iter_mut() {

View file

@ -1,3 +1,4 @@
#![allow(deprecated)]
//! Marker uniqueness test (§4.1, §17.4). //! Marker uniqueness test (§4.1, §17.4).
//! //!
//! Asserts that no `NYX_PWN_*` marker from one cap's corpus is a substring //! Asserts that no `NYX_PWN_*` marker from one cap's corpus is a substring

View file

@ -1,3 +1,4 @@
#![allow(clippy::field_reassign_with_default)]
//! Phase 30 (Track C — security): coverage for //! Phase 30 (Track C — security): coverage for
//! [`crate::dynamic::policy::evaluate`] deny rules. //! [`crate::dynamic::policy::evaluate`] deny rules.
//! //!

View file

@ -142,12 +142,13 @@ fn flask_eval_verdict() -> VerifyResult {
} }
fn flask_eval_sandbox_options() -> SandboxOptions { fn flask_eval_sandbox_options() -> SandboxOptions {
let mut opts = SandboxOptions::default(); SandboxOptions {
opts.backend = SandboxBackend::Docker; backend: SandboxBackend::Docker,
opts.env_passthrough = vec!["NYX_PAYLOAD".into()]; env_passthrough: vec!["NYX_PAYLOAD".into()],
opts.timeout = Duration::from_secs(30); timeout: Duration::from_secs(30),
opts.memory_mib = 256; memory_mib: 256,
opts ..SandboxOptions::default()
}
} }
fn workspace_root() -> PathBuf { fn workspace_root() -> PathBuf {

View file

@ -1,3 +1,4 @@
#![allow(clippy::field_reassign_with_default)]
//! Phase 04 acceptance: callgraph-aware //! Phase 04 acceptance: callgraph-aware
//! [`SpecDerivationStrategy::FromCallgraphEntry`]. //! [`SpecDerivationStrategy::FromCallgraphEntry`].
//! //!

View file

@ -1,3 +1,4 @@
#![allow(clippy::field_reassign_with_default)]
//! Phase 01, Track A.1: integration coverage for //! Phase 01, Track A.1: integration coverage for
//! `HarnessSpec::from_finding_opts` strategy fall-through. //! `HarnessSpec::from_finding_opts` strategy fall-through.
//! //!

View file

@ -27,34 +27,36 @@ use nyx_scanner::patterns::{FindingCategory, Severity};
/// and a synthetic per-name summary, so the framework adapter registry /// and a synthetic per-name summary, so the framework adapter registry
/// resolves a binding when the fixture's source matches an adapter. /// resolves a binding when the fixture's source matches an adapter.
fn make_diag(path: &str, handler: &str, line: usize, cap: Cap, rule_id: &str) -> Diag { fn make_diag(path: &str, handler: &str, line: usize, cap: Cap, rule_id: &str) -> Diag {
let mut ev = Evidence::default(); let ev = Evidence {
ev.flow_steps = vec![ flow_steps: vec![
FlowStep { FlowStep {
step: 0, step: 0,
kind: FlowStepKind::Source, kind: FlowStepKind::Source,
file: path.into(), file: path.into(),
line: line as u32, line: line as u32,
col: 0, col: 0,
snippet: None, snippet: None,
variable: None, variable: None,
callee: None, callee: None,
function: Some(handler.into()), function: Some(handler.into()),
is_cross_file: false, is_cross_file: false,
}, },
FlowStep { FlowStep {
step: 1, step: 1,
kind: FlowStepKind::Sink, kind: FlowStepKind::Sink,
file: path.into(), file: path.into(),
line: line as u32, line: line as u32,
col: 0, col: 0,
snippet: None, snippet: None,
variable: None, variable: None,
callee: None, callee: None,
function: Some(handler.into()), function: Some(handler.into()),
is_cross_file: false, is_cross_file: false,
}, },
]; ],
ev.sink_caps = cap.bits(); sink_caps: cap.bits(),
..Evidence::default()
};
Diag { Diag {
path: path.into(), path: path.into(),
line, line,

View file

@ -48,8 +48,7 @@ fn read_fixture(stub_dir: &str, name: &str) -> String {
/// begin with `//`; the payload is the surviving line. /// begin with `//`; the payload is the surviving line.
fn extract_payload(s: &str) -> String { fn extract_payload(s: &str) -> String {
s.lines() s.lines()
.filter(|l| !l.trim().is_empty() && !l.trim_start().starts_with("//")) .rfind(|l| !l.trim().is_empty() && !l.trim_start().starts_with("//"))
.last()
.unwrap_or("") .unwrap_or("")
.trim() .trim()
.to_owned() .to_owned()