nyx/src/dynamic/lang/php.rs

1389 lines
52 KiB
Rust
Raw Normal View History

//! PHP harness emitter.
//!
//! Phase 15 (Track B PHP vertical) replaces the single legacy `emit`
//! body with dispatch over [`PhpShape`] — the cross product of
//! [`EntryKind`] and a lightweight per-file shape detector that
//! inspects the entry file for Slim/Laravel/Symfony route closures,
//! `$argv`-driven CLI scripts, and top-level script bodies.
//!
//! Each shape emits a single `harness.php` that:
//! 1. Reads the payload from `NYX_PAYLOAD` / `NYX_PAYLOAD_B64` env vars.
//! 2. Includes the entry file (`entry.php`) from the workdir.
//! 3. Invokes the entry function / closure via the per-shape adapter.
//! 4. Catches all Throwables so the harness exit stays observable.
//!
//! Sink-reachability probe: fixtures explicitly emit `__NYX_SINK_HIT__`
//! before the actual sink call (same pattern as Rust / JS fixtures).
//!
//! Payload slot support:
//! - `PayloadSlot::Param(n)` — n-th positional argument.
//! - `PayloadSlot::EnvVar(name)` — set `$_ENV`/`putenv()` before calling.
//! - `PayloadSlot::Stdin` — wrap `STDIN` with the payload.
//! - `PayloadSlot::Argv(n)` — appended to `$argv` for CLI shapes.
//! - `PayloadSlot::QueryParam(name)` — surfaced via `$_GET[name]` /
//! request stub query for route closures.
//! - `PayloadSlot::HttpBody` — surfaced via `$_POST` / request stub body
//! for route closures.
//!
//! Build: no compilation step. Command is `php harness.php`.
//! Build container: `nyx-build-php:{toolchain_id}` (deferred; §19.1).
use crate::dynamic::environment::{Environment, RuntimeArtifacts};
use crate::dynamic::lang::{ChainStepHarness, ChainStepTerminal, HarnessSource, LangEmitter};
use crate::dynamic::spec::{EntryKind, HarnessSpec, PayloadSlot};
use crate::evidence::UnsupportedReason;
use std::path::PathBuf;
/// Zero-sized [`LangEmitter`] handle for PHP. Method bodies delegate to the
/// existing free functions in this module.
pub struct PhpEmitter;
/// Entry kinds the PHP emitter understands after Phase 15.
///
/// `HttpRoute` covers Slim / Laravel / Symfony route closures.
/// `CliSubcommand` covers `$argv`-driven CLI scripts. `Function`
/// covers plain functions and top-level scripts.
const SUPPORTED: &[EntryKind] = &[
EntryKind::Function,
EntryKind::HttpRoute,
EntryKind::CliSubcommand,
];
impl LangEmitter for PhpEmitter {
fn emit(&self, spec: &HarnessSpec) -> Result<HarnessSource, UnsupportedReason> {
emit(spec)
}
fn entry_kinds_supported(&self) -> &'static [EntryKind] {
SUPPORTED
}
fn entry_kind_hint(&self, attempted: EntryKind) -> String {
format!(
"php emitter supports {SUPPORTED:?}; this finding's enclosing context is `EntryKind::{attempted}` — see Phase 15 shape dispatch"
)
}
fn materialize_runtime(&self, env: &Environment) -> RuntimeArtifacts {
materialize_php(env)
}
fn compose_chain_step(
&self,
prev_output: Option<&[u8]>,
terminal: Option<&ChainStepTerminal>,
) -> ChainStepHarness {
chain_step(prev_output, terminal)
}
}
/// Phase 26 — PHP chain-step harness.
///
/// Splices the PHP probe shim ([`probe_shim`]) in front of a minimal
/// driver that reads `NYX_PREV_OUTPUT` via `getenv()` and forwards it
/// on stdout. When the step is the chain's terminal step the driver
/// also calls `__nyx_probe(callee, [prev])` and emits the
/// [`ChainStepHarness::SINK_HIT_SENTINEL`] so the runner flips
/// `sink_hit` for the chain.
fn chain_step(
prev_output: Option<&[u8]>,
terminal: Option<&ChainStepTerminal>,
) -> ChainStepHarness {
let shim = probe_shim();
let mut driver = String::from(
"$prev = getenv(\"NYX_PREV_OUTPUT\");\nif ($prev === false) { $prev = \"\"; }\necho $prev;\n",
);
if let Some(t) = terminal {
let callee = php_string_literal(&t.sink_callee);
let sentinel = php_string_literal(ChainStepHarness::SINK_HIT_SENTINEL);
driver.push_str(&format!(
"__nyx_probe({callee}, [$prev]);\necho \"\\n\" . {sentinel} . \"\\n\";\n",
));
}
let source = format!("<?php\n{shim}\n{driver}");
ChainStepHarness {
source,
filename: "step.php".to_owned(),
command: vec!["php".to_owned(), "step.php".to_owned()],
extra_env: prev_output
.map(|bytes| {
vec![(
ChainStepHarness::PREV_OUTPUT_ENV.to_owned(),
String::from_utf8_lossy(bytes).into_owned(),
)]
})
.unwrap_or_default(),
extra_files: Vec::new(),
}
}
/// Escape a string for safe PHP double-quoted literal embedding.
/// Backslash and double-quote escape only; bytes outside printable
/// ASCII are left to PHP's source decoder.
fn php_string_literal(s: &str) -> String {
let escaped = s.replace('\\', "\\\\").replace('"', "\\\"");
format!("\"{escaped}\"")
}
// ── Phase 15: shape detector ─────────────────────────────────────────────────
/// Concrete per-file shape resolved by reading the entry source.
///
/// One harness template per variant. When the entry file is unreadable
/// or no marker fires the detector defaults to [`PhpShape::Generic`],
/// preserving the pre-Phase-15 behaviour (direct function call).
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum PhpShape {
/// Slim / Laravel / Symfony route closure. Harness builds a
/// minimal request stub (query/body) and invokes the closure
/// resolved from `$GLOBALS['__nyx_route']` (which the entry file
/// publishes during include).
RouteClosure,
/// CLI script driven by `$argv`. Harness mutates `$argv` then
/// includes the entry file (whose top-level body reads `$argv`),
/// or — when the spec names a function — calls the function after
/// setting `$argv`.
CliArgvScript,
/// Top-level script body — no function entry point. Harness just
/// includes the entry file (the include itself runs the body).
TopLevelScript,
/// Plain function — pre-Phase-15 default. Harness calls
/// `funcName($payload)` directly.
Generic,
}
impl PhpShape {
/// Detect the shape from `(spec, source)`. Framework markers in
/// the source win over `spec.entry_kind`.
pub fn detect(spec: &HarnessSpec, source: &str) -> Self {
let entry = spec.entry_name.as_str();
let kind = spec.entry_kind;
let has_route_marker = source.contains("$app->get(")
|| source.contains("$app->post(")
|| source.contains("$app->any(")
|| source.contains("$app->map(")
|| source.contains("$router->get(")
|| source.contains("$router->post(")
|| source.contains("Route::get(")
|| source.contains("Route::post(")
|| source.contains("Route::any(")
|| source.contains("// nyx-shape: route");
let has_argv = source.contains("$argv") || source.contains("// nyx-shape: cli");
let has_function_decl = source.contains("function ")
&& !source.trim_start().starts_with("<?php\n//");
let entry_named_function = entry != "main"
&& entry != "__main__"
&& !entry.is_empty()
&& source.contains(&format!("function {entry}"));
if has_route_marker {
return Self::RouteClosure;
}
if has_argv && !entry_named_function {
return Self::CliArgvScript;
}
if kind == EntryKind::HttpRoute {
return Self::RouteClosure;
}
if kind == EntryKind::CliSubcommand {
return Self::CliArgvScript;
}
// TopLevelScript only fires when we actually saw the source
// and confirmed there's no function declaration to call. When
// the source is unreadable (empty), fall through to Generic so
// the legacy pre-Phase-15 behaviour (direct named-function call)
// survives.
if !source.is_empty() && !has_function_decl && entry.is_empty() {
return Self::TopLevelScript;
}
Self::Generic
}
}
/// Public wrapper to detect the shape for a finalised `HarnessSpec`,
/// reading the entry file from disk.
pub fn detect_shape(spec: &HarnessSpec) -> PhpShape {
let src = read_entry_source(&spec.entry_file);
PhpShape::detect(spec, &src)
}
fn read_entry_source(entry_file: &str) -> String {
let candidates = [PathBuf::from(entry_file), PathBuf::from(".").join(entry_file)];
for path in &candidates {
if let Ok(s) = std::fs::read_to_string(path) {
return s;
}
}
String::new()
}
/// Phase 09 — Track D.2: synthesise a `composer.json` with the captured
/// PHP version pin and (where known) the framework deps.
pub fn materialize_php(env: &Environment) -> RuntimeArtifacts {
let mut artifacts = RuntimeArtifacts::new();
let php_ver = env
.toolchain
.version_string
.split('.')
.take(2)
.collect::<Vec<_>>()
.join(".");
let php_ver = if php_ver.is_empty() {
"8.1".to_owned()
} else {
php_ver
};
let mut body = String::with_capacity(128);
body.push_str("{\n");
body.push_str(" \"name\": \"nyx/harness\",\n");
body.push_str(" \"require\": {\n");
body.push_str(&format!(" \"php\": \">={php_ver}\"\n"));
body.push_str(" }\n");
body.push_str("}\n");
artifacts.push("composer.json", body);
artifacts
}
/// Source of the `__nyx_probe` shim for the PHP harness (Phase 06 —
/// Track C.1).
pub fn probe_shim() -> &'static str {
r#"
// ── __nyx_probe shim (Phase 06 — Track C.1, Phase 08 — Track C.4 + C.5) ──────
const __NYX_DENY_SUBSTRINGS = [
'TOKEN','SECRET','PASSWORD','PASSWD','API_KEY','APIKEY','PRIVATE_KEY',
'CREDENTIAL','SESSION','COOKIE','AUTH','BEARER','AWS_ACCESS','AWS_SESSION',
'GH_TOKEN','GITHUB_TOKEN','NPM_TOKEN','PYPI_TOKEN','DOCKER_PASS',
];
const __NYX_PAYLOAD_LIMIT = 16 * 1024;
const __NYX_REDACTED = '<redacted-by-nyx-policy>';
function __nyx_is_denied_key(string $k): bool {
$ku = strtoupper($k);
foreach (__NYX_DENY_SUBSTRINGS as $n) {
if (strpos($ku, $n) !== false) return true;
}
return false;
}
function __nyx_witness(string $sinkCallee, array $args): array {
$env = [];
foreach ($_ENV as $k => $v) {
$env[(string)$k] = __nyx_is_denied_key((string)$k) ? __NYX_REDACTED : (string)$v;
}
// Sort for deterministic output.
ksort($env);
$payload = (string) (getenv('NYX_PAYLOAD') ?: '');
$pb = substr($payload, 0, __NYX_PAYLOAD_LIMIT);
$bytes = [];
for ($i = 0; $i < strlen($pb); $i++) $bytes[] = ord($pb[$i]);
$repr = [];
foreach ($args as $a) $repr[] = is_string($a) ? $a : (string) $a;
return [
'env_snapshot' => $env,
'cwd' => @getcwd() ?: '',
'payload_bytes' => $bytes,
'callee' => $sinkCallee,
'args_repr' => $repr,
];
}
function __nyx_emit(array $rec): void {
$p = getenv('NYX_PROBE_PATH');
if ($p === false || $p === '') return;
$line = json_encode($rec) . "\n";
@file_put_contents($p, $line, FILE_APPEND);
}
function __nyx_probe(string $sinkCallee, ...$args): void {
$ser = [];
foreach ($args as $a) {
if (is_int($a)) {
$ser[] = ['kind' => 'Int', 'value' => $a];
} else {
$ser[] = ['kind' => 'String', 'value' => (string) $a];
}
}
__nyx_emit([
'sink_callee' => $sinkCallee,
'args' => $ser,
'captured_at_ns' => (int) (microtime(true) * 1e9),
'payload_id' => (string) (getenv('NYX_PAYLOAD_ID') ?: ''),
'kind' => ['kind' => 'Normal'],
'witness' => __nyx_witness($sinkCallee, $args),
]);
}
// Phase 08: PHP cannot catch SIGSEGV from userland, but pcntl_signal and
// register_shutdown_function intercept SIGABRT-class fatal errors.
function __nyx_install_crash_guard(string $sinkCallee): void {
$emit_crash = function (string $signalName) use ($sinkCallee) {
__nyx_emit([
'sink_callee' => $sinkCallee,
'args' => [],
'captured_at_ns' => (int) (microtime(true) * 1e9),
'payload_id' => (string) (getenv('NYX_PAYLOAD_ID') ?: ''),
'kind' => ['kind' => 'Crash', 'signal' => $signalName],
'witness' => __nyx_witness($sinkCallee, []),
]);
};
set_error_handler(function ($errno, $errstr) use ($emit_crash) {
if ($errno & (E_ERROR | E_PARSE | E_CORE_ERROR | E_COMPILE_ERROR | E_USER_ERROR)) {
$emit_crash('SIGABRT');
}
return false;
});
register_shutdown_function(function () use ($emit_crash) {
$err = error_get_last();
if ($err && ($err['type'] & (E_ERROR | E_PARSE | E_CORE_ERROR | E_COMPILE_ERROR))) {
$emit_crash('SIGABRT');
}
});
if (function_exists('pcntl_signal') && function_exists('pcntl_async_signals')) {
pcntl_async_signals(true);
foreach ([SIGABRT, defined('SIGBUS') ? SIGBUS : null, defined('SIGFPE') ? SIGFPE : null, defined('SIGILL') ? SIGILL : null] as $sig) {
if ($sig === null) continue;
pcntl_signal($sig, function ($s) use ($emit_crash) {
$name = 'SIGABRT';
if (defined('SIGABRT') && $s === SIGABRT) $name = 'SIGABRT';
if (defined('SIGBUS') && $s === SIGBUS) $name = 'SIGBUS';
if (defined('SIGFPE') && $s === SIGFPE) $name = 'SIGFPE';
if (defined('SIGILL') && $s === SIGILL) $name = 'SIGILL';
$emit_crash($name);
pcntl_signal($s, SIG_DFL);
posix_kill(posix_getpid(), $s);
});
}
}
}
// Phase 10 (Track D.3) stub helpers. When the verifier spawned a SqlStub it
// publishes the queries-log path through NYX_SQL_LOG; a sink call site that
// wants the host-side stub to see its query appends one record-per-call. The
// helper is a no-op when NYX_SQL_LOG is unset so the same fixture source still
// runs under harness modes that didn't spawn a stub. Mirrors the Python and
// Node shims so the host-side SqlStub log-line format (hash-space-prefixed
// detail lines, then the query line) is identical across language emitters.
function __nyx_stub_sql_record($query, array $detail = []): void {
$p = getenv('NYX_SQL_LOG');
if ($p === false || $p === '') return;
$buf = '';
foreach ($detail as $k => $v) {
$buf .= '# ' . (string)$k . ': ' . (string)$v . "\n";
}
$q = (string)$query;
$buf .= $q;
if (substr($q, -1) !== "\n") $buf .= "\n";
@file_put_contents($p, $buf, FILE_APPEND);
}
// Phase 10 (Track D.3) HTTP recording helper. When the verifier spawned an
// HttpStub it publishes the side-channel log path through NYX_HTTP_LOG; a
// sink call site whose outbound request never reaches the on-the-wire
// listener (DNS-mocked, network-isolated sandbox, pre-flight check) can
// call this helper to surface the attempted call. Format matches the SQL
// helper so the host-side merger parses both streams identically.
function __nyx_stub_http_record($method, $url, $body = null, array $detail = []): void {
$p = getenv('NYX_HTTP_LOG');
if ($p === false || $p === '') return;
$buf = '';
$buf .= '# method: ' . (string)$method . "\n";
$buf .= '# url: ' . (string)$url . "\n";
if ($body !== null) {
$buf .= '# body: ' . (string)$body . "\n";
}
foreach ($detail as $k => $v) {
$buf .= '# ' . (string)$k . ': ' . (string)$v . "\n";
}
$buf .= (string)$method . ' ' . (string)$url . "\n";
@file_put_contents($p, $buf, FILE_APPEND);
}
"#
}
/// Emit a PHP harness for `spec`.
pub fn emit(spec: &HarnessSpec) -> Result<HarnessSource, UnsupportedReason> {
match &spec.payload_slot {
PayloadSlot::Param(_)
| PayloadSlot::EnvVar(_)
| PayloadSlot::Stdin
| PayloadSlot::Argv(_)
| PayloadSlot::QueryParam(_)
| PayloadSlot::HttpBody => {}
}
// Phase 03 (Track J.1): deserialize-sink short-circuit.
if spec.expected_cap == crate::labels::Cap::DESERIALIZE {
return Ok(emit_deserialize_harness(spec));
}
// Phase 04 (Track J.2): SSTI-sink short-circuit.
if spec.expected_cap == crate::labels::Cap::SSTI {
return Ok(emit_ssti_harness(spec));
}
// Phase 05 (Track J.3): XXE-sink short-circuit.
if spec.expected_cap == crate::labels::Cap::XXE {
return Ok(emit_xxe_harness(spec));
}
// Phase 06 (Track J.4): LDAP_INJECTION-sink short-circuit.
if spec.expected_cap == crate::labels::Cap::LDAP_INJECTION {
return Ok(emit_ldap_harness(spec));
}
// Phase 07 (Track J.5): XPATH_INJECTION-sink short-circuit.
if spec.expected_cap == crate::labels::Cap::XPATH_INJECTION {
return Ok(emit_xpath_harness(spec));
}
// Phase 08 (Track J.6): HEADER_INJECTION-sink short-circuit.
if spec.expected_cap == crate::labels::Cap::HEADER_INJECTION {
return Ok(emit_header_injection_harness(spec));
}
// Phase 09 (Track J.7): OPEN_REDIRECT-sink short-circuit.
if spec.expected_cap == crate::labels::Cap::OPEN_REDIRECT {
return Ok(emit_open_redirect_harness(spec));
}
let entry_source = read_entry_source(&spec.entry_file);
let shape = PhpShape::detect(spec, &entry_source);
let source = generate_source(spec, shape);
Ok(HarnessSource {
source,
filename: "harness.php".to_owned(),
command: vec!["php".to_owned(), "harness.php".to_owned()],
extra_files: vec![],
entry_subpath: Some("entry.php".to_owned()),
})
}
/// Phase 03 — Track J.1 deserialize harness for PHP.
///
/// Wraps a call to `unserialize($input, ['allowed_classes' => false])`.
/// The shim parses the payload's `NYX_GADGET_CLASS:<class>` marker;
/// when the marker class is outside the allowlist (`__primitive_int`)
/// the shim writes a [`crate::dynamic::probe::ProbeKind::Deserialize`]
/// probe with `gadget_chain_invoked: true` — simulating the
/// `__wakeup` observer firing on a `__PHP_Incomplete_Class`.
pub fn emit_deserialize_harness(_spec: &HarnessSpec) -> HarnessSource {
let shim = probe_shim();
let body = format!(
r#"<?php
// Nyx dynamic harness — deserialize (Phase 03 / Track J.1).
{shim}
function _nyx_deserialize_probe(bool $invoked): void {{
$p = getenv('NYX_PROBE_PATH');
if ($p === false || $p === '') return;
$rec = [
'sink_callee' => 'unserialize',
'args' => [],
'captured_at_ns' => (int) (hrtime(true)),
'payload_id' => (string) (getenv('NYX_PAYLOAD_ID') ?: ''),
'kind' => ['kind' => 'Deserialize', 'gadget_chain_invoked' => $invoked],
'witness' => __nyx_witness('unserialize', []),
];
@file_put_contents($p, json_encode($rec) . "\n", FILE_APPEND);
}}
$payload = (string) (getenv('NYX_PAYLOAD') ?: '');
$prefix = 'NYX_GADGET_CLASS:';
if (strncmp($payload, $prefix, strlen($prefix)) === 0) {{
$cls = substr($payload, strlen($prefix));
$allowed = ['__primitive_int', '__primitive_string'];
if (!in_array($cls, $allowed, true)) {{
_nyx_deserialize_probe(true);
}}
}}
"#
);
HarnessSource {
source: body,
filename: "harness.php".to_owned(),
command: vec!["php".to_owned(), "harness.php".to_owned()],
extra_files: vec![],
entry_subpath: None,
}
}
/// Phase 04 — Track J.2 SSTI harness for PHP (Twig).
///
/// Reads `NYX_PAYLOAD`, simulates Twig's `{{expr}}` evaluation, prints
/// `{"render": "<result>"}` plus the sink-hit sentinel. Synthetic
/// renderer keeps the corpus deterministic without bundling Twig in
/// the sandbox image.
pub fn emit_ssti_harness(_spec: &HarnessSpec) -> HarnessSource {
let shim = probe_shim();
let body = format!(
r#"<?php
// Nyx dynamic harness — SSTI Twig (Phase 04 / Track J.2).
{shim}
function _nyx_twig_render(string $payload): string {{
return preg_replace_callback('/\{{\{{(.+?)\}}\}}/', function ($m) {{
$expr = trim($m[1]);
if (preg_match('/^(\d+)\s*\*\s*(\d+)$/', $expr, $mm)) {{
return (string) ((int) $mm[1] * (int) $mm[2]);
}}
if (preg_match('/^(\d+)\s*\+\s*(\d+)$/', $expr, $mm)) {{
return (string) ((int) $mm[1] + (int) $mm[2]);
}}
return $m[0];
}}, $payload) ?? $payload;
}}
function _nyx_ssti_probe(string $rendered): void {{
$p = getenv('NYX_PROBE_PATH');
if ($p === false || $p === '') return;
$rec = [
'sink_callee' => 'Twig\\Environment::render',
'args' => [['kind' => 'String', 'value' => $rendered]],
'captured_at_ns' => (int) hrtime(true),
'payload_id' => (string) (getenv('NYX_PAYLOAD_ID') ?: ''),
'kind' => ['kind' => 'Normal'],
'witness' => __nyx_witness('Twig\\Environment::render', [$rendered]),
];
@file_put_contents($p, json_encode($rec) . "\n", FILE_APPEND);
}}
$payload = (string) (getenv('NYX_PAYLOAD') ?: '');
$rendered = _nyx_twig_render($payload);
_nyx_ssti_probe($rendered);
echo "__NYX_SINK_HIT__\n";
echo json_encode(["render" => $rendered]) . "\n";
"#
);
HarnessSource {
source: body,
filename: "harness.php".to_owned(),
command: vec!["php".to_owned(), "harness.php".to_owned()],
extra_files: vec![],
entry_subpath: None,
}
}
/// 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">`
/// declarations, expands them inside `&name;` element references
/// (matching `simplexml_load_string` / `DOMDocument` with the entity
/// loader re-enabled), and writes a `ProbeKind::Xxe` probe whose
/// `entity_expanded` flag tracks whether the substitution fired.
pub fn emit_xxe_harness(_spec: &HarnessSpec) -> HarnessSource {
let shim = probe_shim();
let body = format!(
r#"<?php
// Nyx dynamic harness — XXE simplexml_load_string (Phase 05 / Track J.3).
{shim}
function _nyx_libxml_parse(string $payload): array {{
$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;
$rendered = preg_replace_callback('/&(\w+);/', function ($m) use ($entities, &$expanded) {{
if (array_key_exists($m[1], $entities)) {{
$expanded = true;
return $entities[$m[1]];
}}
return $m[0];
}}, $payload) ?? $payload;
return [$rendered, $expanded];
}}
function _nyx_xxe_probe(string $rendered, bool $expanded): void {{
$p = getenv('NYX_PROBE_PATH');
if ($p === false || $p === '') return;
$rec = [
'sink_callee' => 'simplexml_load_string',
'args' => [['kind' => 'String', 'value' => $rendered]],
'captured_at_ns' => (int) hrtime(true),
'payload_id' => (string) (getenv('NYX_PAYLOAD_ID') ?: ''),
'kind' => ['kind' => 'Xxe', 'entity_expanded' => $expanded],
'witness' => __nyx_witness('simplexml_load_string', [$rendered]),
];
@file_put_contents($p, json_encode($rec) . "\n", FILE_APPEND);
}}
$payload = (string) (getenv('NYX_PAYLOAD') ?: '');
[$rendered, $expanded] = _nyx_libxml_parse($payload);
_nyx_xxe_probe($rendered, $expanded);
echo "__NYX_SINK_HIT__\n";
echo json_encode(["render" => $rendered, "entity_expanded" => $expanded]) . "\n";
"#
);
HarnessSource {
source: body,
filename: "harness.php".to_owned(),
command: vec!["php".to_owned(), "harness.php".to_owned()],
extra_files: vec![],
entry_subpath: None,
}
}
/// Phase 06 — Track J.4 LDAP-injection harness for PHP (`ldap_search`).
///
/// Reads `NYX_PAYLOAD`, splices it into a `(uid=<payload>)` filter,
/// evaluates the filter against the in-sandbox LDAP directory (three
/// users: `alice`, `bob`, `carol`) using the same RFC-4515 subset the
/// [`crate::dynamic::stubs::ldap_server`] stub implements, and writes
/// a `ProbeKind::Ldap { entries_returned }` probe whose `n` is the
/// count the directory returned. Mirrors the synthetic-harness
/// pattern used by Phase 03 / 04 / 05.
pub fn emit_ldap_harness(_spec: &HarnessSpec) -> HarnessSource {
let shim = probe_shim();
let body = format!(
r#"<?php
// Nyx dynamic harness — LDAP_INJECTION ldap_search (Phase 06 / Track J.4).
{shim}
$NYX_LDAP_USERS = ['alice', 'bob', 'carol'];
function _nyx_attr_match(string $pattern, string $uid): bool {{
if ($pattern === '*') return true;
$star = strpos($pattern, '*');
if ($star === false) return $pattern === $uid;
$prefix = substr($pattern, 0, $star);
$suffix = substr($pattern, $star + 1);
return str_starts_with($uid, $prefix) && str_ends_with($uid, $suffix);
}}
function _nyx_split_clauses(string $src): array {{
$out = [];
$i = 0;
$n = strlen($src);
while ($i < $n) {{
if ($src[$i] !== '(') {{ $i++; continue; }}
$depth = 0;
$start = $i;
while ($i < $n) {{
$c = $src[$i];
if ($c === '(') $depth++;
elseif ($c === ')') {{
$depth--;
if ($depth === 0) {{ $i++; break; }}
}}
$i++;
}}
$out[] = substr($src, $start, $i - $start);
}}
return $out;
}}
function _nyx_inner_has_break(string $inner): bool {{
$depth = 0;
$n = strlen($inner);
for ($i = 0; $i < $n; $i++) {{
$c = $inner[$i];
if ($c === '(') $depth++;
elseif ($c === ')') {{
$depth--;
if ($depth < 0) return true;
}}
}}
return false;
}}
function _nyx_match_one(string $filt, string $uid): bool {{
$f = trim($filt);
if (!(str_starts_with($f, '(') && str_ends_with($f, ')'))) return true;
$inner = substr($f, 1, strlen($f) - 2);
if (_nyx_inner_has_break($inner)) return true;
if (str_starts_with($inner, '&') || str_starts_with($inner, '|')) {{
$clauses = _nyx_split_clauses(substr($inner, 1));
if (empty($clauses)) return false;
$is_and = str_starts_with($inner, '&');
$ok = $is_and;
foreach ($clauses as $c) {{
$m = _nyx_match_one($c, $uid);
$ok = $is_and ? ($ok && $m) : ($ok || $m);
}}
return $ok;
}}
$eq = strpos($inner, '=');
if ($eq === false) return true;
$attr = strtolower(substr($inner, 0, $eq));
$pattern = substr($inner, $eq + 1);
if ($attr !== 'uid' && $attr !== 'cn') return true;
return _nyx_attr_match($pattern, $uid);
}}
function _nyx_ldap_count(string $filt, array $users): int {{
$f = trim($filt);
if ($f === '') return 0;
if (!(str_starts_with($f, '(') && str_ends_with($f, ')'))) return count($users);
$inner = substr($f, 1, strlen($f) - 2);
if (_nyx_inner_has_break($inner)) return count($users);
$count = 0;
foreach ($users as $u) {{
if (_nyx_match_one($f, $u)) $count++;
}}
return $count;
}}
function _nyx_ldap_probe(string $filt, int $entries_returned): void {{
$p = getenv('NYX_PROBE_PATH');
if ($p === false || $p === '') return;
$rec = [
'sink_callee' => 'ldap_search',
'args' => [['kind' => 'String', 'value' => $filt]],
'captured_at_ns' => (int) hrtime(true),
'payload_id' => (string) (getenv('NYX_PAYLOAD_ID') ?: ''),
'kind' => ['kind' => 'Ldap', 'entries_returned' => $entries_returned],
'witness' => __nyx_witness('ldap_search', [$filt]),
];
@file_put_contents($p, json_encode($rec) . "\n", FILE_APPEND);
}}
$payload = (string) (getenv('NYX_PAYLOAD') ?: '');
$filt = '(uid=' . $payload . ')';
$count = _nyx_ldap_count($filt, $NYX_LDAP_USERS);
_nyx_ldap_probe($filt, $count);
echo "__NYX_SINK_HIT__\n";
echo json_encode(['filter' => $filt, 'entries_returned' => $count]) . "\n";
"#
);
HarnessSource {
source: body,
filename: "harness.php".to_owned(),
command: vec!["php".to_owned(), "harness.php".to_owned()],
extra_files: vec![],
entry_subpath: None,
}
}
/// Phase 07 — Track J.5 XPath-injection harness for PHP
/// (`DOMXPath::query`).
///
/// Reads `NYX_PAYLOAD`, splices it into a `//user[@name='<payload>']`
/// expression, evaluates the resulting expression against the
/// canonical XML staged in the workdir via
/// [`crate::dynamic::stubs::xpath_document`] (three `<user>`
/// records), and writes a `ProbeKind::Xpath { nodes_returned }`
/// probe whose `n` is the count the evaluator returned. Mirrors the
/// synthetic-harness pattern used by Phase 03 / 04 / 05 / 06; a
/// future structural fix will link real `DOMXPath` via the staged
/// document.
pub fn emit_xpath_harness(_spec: &HarnessSpec) -> HarnessSource {
let shim = probe_shim();
let corpus_filename = crate::dynamic::stubs::xpath_document::XPATH_CORPUS_FILENAME;
let corpus_xml = crate::dynamic::stubs::xpath_document::XPATH_CORPUS_XML;
let body = format!(
r#"<?php
// Nyx dynamic harness — XPATH_INJECTION DOMXPath::query (Phase 07 / Track J.5).
{shim}
// Synthetic in-process XPath evaluator over the canonical staged
// document — counts <user> nodes that satisfy the `[@name='…']`
// predicate the host code synthesised from the payload. Real
// `DOMXPath::query` is not invoked (the harness ignores `_spec` and
// inlines the evaluator); the differential rule still holds because
// the vuln payload's `' or '1'='1` tail rewraps the selector into a
// match-everything shape.
$NYX_XPATH_USERS = ['alice', 'bob', 'carol'];
function _nyx_xpath_select($expr, array $users): int {{
// Recognise the canonical `//user[@name='<payload>']` shape the
// synthetic harness emits. Anything else falls through to "no
// match" so a malformed expression cannot accidentally confirm.
$needle = "//user[@name=";
if (strncmp($expr, $needle, strlen($needle)) !== 0) {{
return 0;
}}
$rest = substr($expr, strlen($needle));
if (!str_ends_with($rest, ']')) {{
return 0;
}}
$predicate = substr($rest, 0, strlen($rest) - 1);
if (preg_match("/^'([^']*)'(.*)\$/", $predicate, $m)) {{
// `name='alice'` → exact-match against the literal
// `name='alice' or '1'='1'` → OR-tail breakouts; presence of
// ` or ` after the closing quote means the selector is now
// tautological → every user matches.
$literal = $m[1];
$tail = trim($m[2]);
if ($tail === '' || $tail === ']') {{
$count = 0;
foreach ($users as $u) {{
if ($u === $literal) $count++;
}}
return $count;
}}
if (preg_match("/^or\\s+/i", $tail)) {{
return count($users);
}}
}}
if (preg_match('/^"([^"]*)"\\s*$/', $predicate, $m)) {{
$literal = $m[1];
$count = 0;
foreach ($users as $u) {{
if ($u === $literal) $count++;
}}
return $count;
}}
if (preg_match("/^concat\\(/i", $predicate)) {{
// `concat('a',\"'\",'b')` benign-escape path: extract the
// joined literal and match exactly once.
if (preg_match_all("/'([^']*)'/", $predicate, $parts)) {{
$joined = '';
foreach ($parts[1] as $p) {{
if ($p === ',"') continue;
$joined .= $p;
}}
// Normalise embedded single-quote literals back to the
// raw character so a `concat`-quoted username collapses
// to the same literal the user typed.
$joined = str_replace(",\"'\",", "'", $joined);
$count = 0;
foreach ($users as $u) {{
if ($u === $joined) $count++;
}}
return $count;
}}
}}
return count($users);
}}
function _nyx_xpath_probe(string $expr, int $nodes_returned): void {{
$p = getenv('NYX_PROBE_PATH');
if ($p === false || $p === '') return;
$rec = [
'sink_callee' => 'DOMXPath::query',
'args' => [['kind' => 'String', 'value' => $expr]],
'captured_at_ns' => (int) hrtime(true),
'payload_id' => (string) (getenv('NYX_PAYLOAD_ID') ?: ''),
'kind' => ['kind' => 'Xpath', 'nodes_returned' => $nodes_returned],
'witness' => __nyx_witness('DOMXPath::query', [$expr]),
];
@file_put_contents($p, json_encode($rec) . "\n", FILE_APPEND);
}}
$payload = (string) (getenv('NYX_PAYLOAD') ?: '');
$expr = "//user[@name='" . $payload . "']";
$nodes = _nyx_xpath_select($expr, $NYX_XPATH_USERS);
_nyx_xpath_probe($expr, $nodes);
echo "__NYX_SINK_HIT__\n";
echo json_encode(['expr' => $expr, 'nodes_returned' => $nodes]) . "\n";
"#
);
let extra_files = vec![(corpus_filename.to_owned(), corpus_xml.to_owned())];
HarnessSource {
source: body,
filename: "harness.php".to_owned(),
command: vec!["php".to_owned(), "harness.php".to_owned()],
extra_files,
entry_subpath: None,
}
}
/// Phase 08 — Track J.6 header-injection harness for PHP (`header()`).
///
/// Reads `NYX_PAYLOAD`, calls a synthetic instrumented `header()`
/// shim that records the *unmodified* value bytes (including any
/// embedded `\r\n`) via a `ProbeKind::HeaderEmit` probe. Mirrors
/// the synthetic-harness pattern used by Phase 03 / 04 / 05 / 06 /
/// 07.
pub fn emit_header_injection_harness(_spec: &HarnessSpec) -> HarnessSource {
let shim = probe_shim();
let body = format!(
r#"<?php
// Nyx dynamic harness — HEADER_INJECTION header() (Phase 08 / Track J.6).
{shim}
function _nyx_header_probe(string $name, string $value): void {{
$p = getenv('NYX_PROBE_PATH');
if ($p === false || $p === '') return;
$rec = [
'sink_callee' => 'header()',
'args' => [
['kind' => 'String', 'value' => $name],
['kind' => 'String', 'value' => $value],
],
'captured_at_ns' => (int) hrtime(true),
'payload_id' => (string) (getenv('NYX_PAYLOAD_ID') ?: ''),
'kind' => ['kind' => 'HeaderEmit', 'name' => $name, 'value' => $value],
'witness' => __nyx_witness('header()', [$name, $value]),
];
@file_put_contents($p, json_encode($rec) . "\n", FILE_APPEND);
}}
$payload = (string) (getenv('NYX_PAYLOAD') ?: '');
$name = 'Set-Cookie';
$value = $payload;
_nyx_header_probe($name, $value);
echo "__NYX_SINK_HIT__\n";
echo json_encode(['name' => $name, 'value' => $value]) . "\n";
"#
);
HarnessSource {
source: body,
filename: "harness.php".to_owned(),
command: vec!["php".to_owned(), "harness.php".to_owned()],
extra_files: Vec::new(),
entry_subpath: None,
}
}
/// Phase 09 — Track J.7 open-redirect harness for PHP (`header("Location: …")` /
/// `Response::redirect`).
///
/// Reads `NYX_PAYLOAD`, calls a synthetic instrumented redirect shim
/// that records the bound `Location:` value plus the request's origin
/// host via a `ProbeKind::Redirect` probe. Mirrors the
/// synthetic-harness pattern used by Phase 03 / 04 / 05 / 06 / 07 / 08.
pub fn emit_open_redirect_harness(_spec: &HarnessSpec) -> HarnessSource {
let shim = probe_shim();
let body = format!(
r#"<?php
// Nyx dynamic harness — OPEN_REDIRECT Response::redirect (Phase 09 / Track J.7).
{shim}
function _nyx_redirect_probe(string $location, string $requestHost): void {{
$p = getenv('NYX_PROBE_PATH');
if ($p === false || $p === '') return;
$rec = [
'sink_callee' => 'Response::redirect',
'args' => [
['kind' => 'String', 'value' => $location],
],
'captured_at_ns' => (int) hrtime(true),
'payload_id' => (string) (getenv('NYX_PAYLOAD_ID') ?: ''),
'kind' => [
'kind' => 'Redirect',
'location' => $location,
'request_host' => $requestHost,
],
'witness' => __nyx_witness('Response::redirect', [$location]),
];
@file_put_contents($p, json_encode($rec) . "\n", FILE_APPEND);
}}
$payload = (string) (getenv('NYX_PAYLOAD') ?: '');
$requestHost = 'example.com';
$location = $payload;
_nyx_redirect_probe($location, $requestHost);
echo "__NYX_SINK_HIT__\n";
echo json_encode(['location' => $location, 'request_host' => $requestHost]) . "\n";
"#
);
HarnessSource {
source: body,
filename: "harness.php".to_owned(),
command: vec!["php".to_owned(), "harness.php".to_owned()],
extra_files: Vec::new(),
entry_subpath: None,
}
}
fn generate_source(spec: &HarnessSpec, shape: PhpShape) -> String {
let entry_fn = &spec.entry_name;
let pre_call = build_pre_call(spec, shape);
let entry_block = build_entry_block(shape);
let call_expr = build_call_expr(spec, shape, entry_fn);
let shim = probe_shim();
let crash_callee = if entry_fn.is_empty() { "main" } else { entry_fn.as_str() };
format!(
r#"<?php
// Nyx dynamic harness — auto-generated, do not edit (Phase 15 — PhpShape::{shape:?}).
{shim}
// ── Payload loading ────────────────────────────────────────────────────────────
function nyx_payload(): string {{
$v = getenv('NYX_PAYLOAD');
if ($v !== false && $v !== '') {{
return $v;
}}
$b64 = getenv('NYX_PAYLOAD_B64');
if ($b64 !== false && $b64 !== '') {{
return base64_decode($b64, true) ?: '';
}}
return '';
}}
$payload = nyx_payload();
// Phase 08 sink-site signal handler: install AFTER payload decode so a crash
// inside `nyx_payload` writes no Crash probe and routes the verifier to
// `Inconclusive(UnrelatedCrash)`. A fatal-error inside the entry call below
// DOES fire the handler and writes a Crash probe to `NYX_PROBE_PATH`.
__nyx_install_crash_guard('{crash_callee}');
// ── Pre-call setup ─────────────────────────────────────────────────────────────
{pre_call}
// ── Entry include ─────────────────────────────────────────────────────────────
{entry_block}
// ── Call entry point ──────────────────────────────────────────────────────────
try {{
$result = {call_expr};
if ($result !== null) {{
echo $result . "\n";
}}
}} catch (Throwable $e) {{
fwrite(STDERR, 'NYX_EXCEPTION: ' . get_class($e) . ': ' . $e->getMessage() . "\n");
}}
"#,
shape = shape,
pre_call = pre_call,
entry_block = entry_block,
call_expr = call_expr,
shim = shim,
crash_callee = crash_callee,
)
}
fn build_pre_call(spec: &HarnessSpec, shape: PhpShape) -> String {
let mut out = String::new();
match &spec.payload_slot {
PayloadSlot::EnvVar(name) => {
out.push_str(&format!(
"putenv({name:?} . '=' . $payload);\n$_ENV[{name:?}] = $payload;\n"
));
}
PayloadSlot::Stdin => {
out.push_str(
"if (defined('STDIN')) {\n $stream = fopen('php://memory', 'r+');\n fwrite($stream, $payload);\n rewind($stream);\n}\n",
);
}
PayloadSlot::Argv(n) => {
out.push_str("$argv = $argv ?? [];\n");
out.push_str("$argv[0] = $argv[0] ?? 'nyx_harness';\n");
for _ in 0..*n {
out.push_str("$argv[] = '';\n");
}
out.push_str("$argv[] = $payload;\n");
out.push_str("$argc = count($argv);\n");
out.push_str("$_SERVER['argv'] = $argv;\n");
out.push_str("$_SERVER['argc'] = $argc;\n");
}
PayloadSlot::QueryParam(name) => {
out.push_str(&format!("$_GET[{name:?}] = $payload;\n"));
out.push_str("$_REQUEST = array_merge($_REQUEST ?? [], $_GET);\n");
}
PayloadSlot::HttpBody => {
out.push_str("$_POST['body'] = $payload;\n");
out.push_str("$GLOBALS['__nyx_body'] = $payload;\n");
}
_ => {}
}
if matches!(shape, PhpShape::CliArgvScript)
&& !matches!(&spec.payload_slot, PayloadSlot::Argv(_))
{
out.push_str("$argv = $argv ?? ['nyx_harness'];\n");
out.push_str("$argv[] = $payload;\n");
out.push_str("$argc = count($argv);\n");
out.push_str("$_SERVER['argv'] = $argv;\n");
out.push_str("$_SERVER['argc'] = $argc;\n");
}
out
}
fn build_entry_block(_shape: PhpShape) -> String {
r#"try {
require_once __DIR__ . '/entry.php';
} catch (Throwable $e) {
fwrite(STDERR, 'NYX_IMPORT_ERROR: ' . $e->getMessage() . "\n");
exit(77);
}"#
.to_owned()
}
fn build_call_expr(spec: &HarnessSpec, shape: PhpShape, func: &str) -> String {
match shape {
PhpShape::TopLevelScript => "null".to_owned(),
PhpShape::CliArgvScript => {
if func.is_empty() || func == "main" || func == "__main__" {
"null".to_owned()
} else if function_exists_call(func) {
format!("{func}()")
} else {
"null".to_owned()
}
}
PhpShape::RouteClosure => {
// Entry script publishes the route closure via
// `$GLOBALS['__nyx_route']`. When the global is missing,
// fall back to calling the named function directly.
format!(
"(isset($GLOBALS['__nyx_route']) && is_callable($GLOBALS['__nyx_route'])) ? call_user_func($GLOBALS['__nyx_route'], $payload) : (function_exists({func:?}) ? {func}($payload) : null)"
)
}
PhpShape::Generic => build_generic_call(spec, func),
}
}
fn build_generic_call(spec: &HarnessSpec, func: &str) -> String {
match &spec.payload_slot {
PayloadSlot::Param(idx) => {
if *idx == 0 {
format!("{func}($payload)")
} else {
let pads = (0..*idx).map(|_| "''").collect::<Vec<_>>().join(", ");
format!("{func}({pads}, $payload)")
}
}
PayloadSlot::EnvVar(_) | PayloadSlot::Stdin => format!("{func}()"),
_ => format!("{func}($payload)"),
}
}
/// Wrap the named-function call in a `function_exists` guard for shapes
/// where the entry function may be optional (CLI scripts whose body is
/// the entry, not a named function).
fn function_exists_call(_func: &str) -> bool {
true
}
#[cfg(test)]
mod tests {
use super::*;
use crate::dynamic::spec::{EntryKind, HarnessSpec, PayloadSlot};
use crate::labels::Cap;
use crate::symbol::Lang;
fn make_spec(payload_slot: PayloadSlot) -> HarnessSpec {
HarnessSpec {
finding_id: "php0000000000001".into(),
entry_file: "src/login.php".into(),
entry_name: "login".into(),
entry_kind: EntryKind::Function,
lang: Lang::Php,
toolchain_id: "php-8".into(),
payload_slot,
expected_cap: Cap::SQL_QUERY,
constraint_hints: vec![],
sink_file: "src/login.php".into(),
sink_line: 10,
spec_hash: "php0000000000001".into(),
derivation: crate::dynamic::spec::SpecDerivationStrategy::FromFlowSteps,
stubs_required: vec![],
framework: None,
java_toolchain: crate::dynamic::spec::JavaToolchain::default(),
}
}
#[test]
fn emit_produces_source() {
let spec = make_spec(PayloadSlot::Param(0));
let harness = emit(&spec).unwrap();
assert!(harness.source.starts_with("<?php"));
assert!(harness.source.contains("NYX_PAYLOAD"));
assert!(harness.source.contains("require_once"));
assert!(harness.source.contains("login($payload)"));
assert_eq!(harness.filename, "harness.php");
assert_eq!(harness.command, vec!["php", "harness.php"]);
}
#[test]
fn emit_param_index_0() {
let spec = make_spec(PayloadSlot::Param(0));
let harness = emit(&spec).unwrap();
assert!(harness.source.contains("login($payload)"));
}
#[test]
fn emit_param_index_2() {
let spec = make_spec(PayloadSlot::Param(2));
let harness = emit(&spec).unwrap();
assert!(harness.source.contains("login('', '', $payload)"));
}
#[test]
fn emit_env_var_slot() {
let spec = make_spec(PayloadSlot::EnvVar("DB_HOST".into()));
let harness = emit(&spec).unwrap();
assert!(harness.source.contains("putenv"));
assert!(harness.source.contains("\"DB_HOST\""));
}
#[test]
fn emit_http_body_now_supported_for_route_shape() {
let mut spec = make_spec(PayloadSlot::HttpBody);
spec.entry_kind = EntryKind::HttpRoute;
let h = emit(&spec).unwrap();
assert!(h.source.contains("$GLOBALS['__nyx_body']"));
}
#[test]
fn emit_entry_subpath_is_entry_php() {
let spec = make_spec(PayloadSlot::Param(0));
let harness = emit(&spec).unwrap();
assert_eq!(harness.entry_subpath, Some("entry.php".to_owned()));
}
#[test]
fn entry_kinds_supported_is_non_empty() {
assert!(!PhpEmitter.entry_kinds_supported().is_empty());
assert!(PhpEmitter
.entry_kinds_supported()
.contains(&EntryKind::Function));
assert!(PhpEmitter
.entry_kinds_supported()
.contains(&EntryKind::HttpRoute));
assert!(PhpEmitter
.entry_kinds_supported()
.contains(&EntryKind::CliSubcommand));
}
#[test]
fn entry_kind_hint_names_attempted_and_phase() {
let hint = PhpEmitter.entry_kind_hint(EntryKind::LibraryApi);
assert!(hint.contains("LibraryApi"));
assert!(hint.contains("Phase 15"));
}
#[test]
fn harness_has_base64_decode() {
let spec = make_spec(PayloadSlot::Param(0));
let harness = emit(&spec).unwrap();
assert!(harness.source.contains("base64_decode"));
assert!(harness.source.contains("NYX_PAYLOAD_B64"));
}
// ── Phase 15: shape detection ────────────────────────────────────────────
fn make_spec_with(kind: EntryKind, name: &str, entry_file: &str) -> HarnessSpec {
let mut s = make_spec(PayloadSlot::Param(0));
s.entry_kind = kind;
s.entry_name = name.to_owned();
s.entry_file = entry_file.to_owned();
s
}
#[test]
fn shape_detect_slim_route_closure() {
let src = "<?php\n$app->get('/run', function ($req, $res) {\n return 'hi';\n});\n";
let spec = make_spec_with(EntryKind::HttpRoute, "run", "entry.php");
assert_eq!(PhpShape::detect(&spec, src), PhpShape::RouteClosure);
}
#[test]
fn shape_detect_laravel_route_closure() {
let src = "<?php\nRoute::get('/run', function ($payload) { return $payload; });\n";
let spec = make_spec_with(EntryKind::HttpRoute, "run", "entry.php");
assert_eq!(PhpShape::detect(&spec, src), PhpShape::RouteClosure);
}
#[test]
fn shape_detect_cli_argv_script() {
let src = "<?php\n$cmd = $argv[1] ?? '';\necho $cmd;\n";
let spec = make_spec_with(EntryKind::CliSubcommand, "main", "entry.php");
assert_eq!(PhpShape::detect(&spec, src), PhpShape::CliArgvScript);
}
#[test]
fn shape_detect_top_level_script() {
let src = "<?php\necho 'hello';\n";
let spec = make_spec_with(EntryKind::Function, "", "entry.php");
assert_eq!(PhpShape::detect(&spec, src), PhpShape::TopLevelScript);
}
#[test]
fn shape_detect_generic_function() {
let src = "<?php\nfunction login($payload) { return $payload; }\n";
let spec = make_spec_with(EntryKind::Function, "login", "entry.php");
assert_eq!(PhpShape::detect(&spec, src), PhpShape::Generic);
}
#[test]
fn route_shape_emits_globals_dispatch() {
let spec = make_spec_with(EntryKind::HttpRoute, "ping", "entry.php");
let src = generate_source(&spec, PhpShape::RouteClosure);
assert!(src.contains("$GLOBALS['__nyx_route']"));
}
#[test]
fn cli_shape_appends_payload_to_argv() {
let spec = make_spec_with(EntryKind::CliSubcommand, "main", "entry.php");
let src = generate_source(&spec, PhpShape::CliArgvScript);
assert!(src.contains("$argv"));
assert!(src.contains("$_SERVER['argv']"));
}
#[test]
fn top_level_script_only_includes() {
let spec = make_spec_with(EntryKind::Function, "", "entry.php");
let src = generate_source(&spec, PhpShape::TopLevelScript);
assert!(src.contains("require_once"));
assert!(src.contains("$result = null"));
}
#[test]
fn emit_splices_probe_shim_and_installs_crash_guard() {
let spec = make_spec(PayloadSlot::Param(0));
let h = emit(&spec).unwrap();
assert!(
h.source.contains("__nyx_probe shim (Phase 06 — Track C.1"),
"probe_shim banner missing from generated harness.php — splicing regressed",
);
assert!(
h.source
.contains("function __nyx_install_crash_guard(string $sinkCallee)"),
"install_crash_guard definition missing from generated harness.php",
);
assert!(
h.source.contains("__nyx_install_crash_guard('login');"),
"install_crash_guard call site missing or wrong callee in harness body",
);
let install_pos = h
.source
.find("__nyx_install_crash_guard('login');")
.unwrap();
let payload_pos = h.source.find("$payload = nyx_payload();").unwrap();
let invoke_pos = h.source.find("login($payload)").unwrap();
assert!(
payload_pos < install_pos && install_pos < invoke_pos,
"install_crash_guard ordering wrong: payload_pos={payload_pos} install_pos={install_pos} invoke_pos={invoke_pos}",
);
}
#[test]
fn probe_shim_publishes_stub_sql_recorder() {
let shim = probe_shim();
assert!(
shim.contains("function __nyx_stub_sql_record"),
"PHP probe shim must define __nyx_stub_sql_record"
);
assert!(
shim.contains("NYX_SQL_LOG"),
"stub recorder must read NYX_SQL_LOG"
);
}
#[test]
fn probe_shim_publishes_stub_http_recorder() {
let shim = probe_shim();
assert!(
shim.contains("function __nyx_stub_http_record"),
"PHP probe shim must define __nyx_stub_http_record"
);
assert!(
shim.contains("NYX_HTTP_LOG"),
"stub recorder must read NYX_HTTP_LOG"
);
}
#[test]
fn chain_step_splices_probe_shim_for_composite_reverify() {
let step = chain_step(Some(b"<prev>"), None);
assert!(
step.source.contains("__nyx_probe"),
"PHP chain step must splice the probe shim"
);
assert!(
step.source.starts_with("<?php"),
"PHP chain step must open with <?php"
);
assert!(
step.source.contains("getenv(\"NYX_PREV_OUTPUT\")"),
"PHP chain step must keep its NYX_PREV_OUTPUT forwarder"
);
let shim_pos = step.source.find("__nyx_probe").unwrap();
let driver_pos = step.source.find("getenv(\"NYX_PREV_OUTPUT\")").unwrap();
assert!(
shim_pos < driver_pos,
"probe shim must come before the driver so the shim's helpers are in scope when a sink rewrite splices in"
);
}
}