//! End-to-end integration tests for the Phase-6 type-hierarchy index //! installation and runtime fan-out wiring. //! //! These tests run the production pass-1 extraction pipeline //! (`extract_all_summaries_from_bytes` + `merge_summaries` + //! `insert_ssa`) on synthetic multi-file sources, then exercise the //! `GlobalSummaries::install_hierarchy` + `resolve_callee_widened` //! contract that the taint engine's runtime callee resolver consumes. //! //! Why integration-level coverage matters //! ────────────────────────────────────── //! The unit tests in `src/summary/tests.rs::hierarchy_widened_tests` //! cover the lookup contract on hand-crafted summaries. These tests //! cover the *upstream* invariant: that pass-1's //! `cfg/hierarchy.rs` extractor populates `FuncSummary::hierarchy_edges` //! correctly for every supported language (Java, Rust, TypeScript, //! Python), and that `install_hierarchy` rebuilds the index off those //! edges so the runtime resolver sees concrete implementers at virtual- //! dispatch call sites. //! //! When this file fails, the gap is somewhere between the AST //! extractor (`src/cfg/hierarchy.rs`), the summary plumbing //! (`FuncSummary::hierarchy_edges`), and the runtime install (`scan.rs` //! `install_hierarchy` call site). Unit-test failures localise more //! tightly; integration-test failures point at the seam. mod common; use nyx_scanner::ast::extract_all_summaries_from_bytes; use nyx_scanner::summary::{CalleeQuery, GlobalSummaries, ssa_summary::SsaFuncSummary}; use nyx_scanner::symbol::{FuncKey, Lang}; use nyx_scanner::utils::config::AnalysisMode; use std::path::Path; use common::test_config; struct File<'a> { namespace: &'a str, bytes: &'a [u8], } /// Run pass-1 extraction + merge over a synthetic file set, then /// install the hierarchy index, mirroring exactly what production /// scan paths do before pass 2 runs. fn build_gs(files: &[File<'_>]) -> GlobalSummaries { let cfg = test_config(AnalysisMode::Taint); let mut all_func: Vec = Vec::new(); let mut all_ssa: Vec<(FuncKey, SsaFuncSummary)> = Vec::new(); for f in files { let path = Path::new(f.namespace); let (func, ssa, _bodies, _auth, _cpi) = extract_all_summaries_from_bytes(f.bytes, path, &cfg, None) .expect("extract_all_summaries_from_bytes must succeed"); all_func.extend(func); all_ssa.extend(ssa); } let mut gs = nyx_scanner::summary::merge_summaries(all_func, None); for (k, s) in all_ssa { gs.insert_ssa(k, s); } gs.install_hierarchy(); gs } // ───────────────────────────────────────────────────────────────────────── // C1, Java interface fan-out // ───────────────────────────────────────────────────────────────────────── /// Pass-1 must extract the `class FileLogger implements ILogger` /// edge. After `install_hierarchy`, a query with /// `receiver_type = ILogger` widens to both ILogger's own method and /// every implementer's overriding method. #[test] fn java_interface_with_two_impls_fans_out_to_both() { // Three files: one defines the interface, two define impls. Each // impl declares `void log(String s)` so the leaf-name lookup has // material to fan out to. let logger_iface = br#" package app; public interface ILogger { void log(String s); } "#; let console_logger = br#" package app; public class ConsoleLogger implements ILogger { public void log(String s) { System.out.println(s); } } "#; let file_logger = br#" package app; public class FileLogger implements ILogger { public void log(String s) { java.io.File f = new java.io.File("/tmp/" + s); } } "#; let gs = build_gs(&[ File { namespace: "src/ILogger.java", bytes: logger_iface, }, File { namespace: "src/ConsoleLogger.java", bytes: console_logger, }, File { namespace: "src/FileLogger.java", bytes: file_logger, }, ]); let h = gs.hierarchy().expect("hierarchy must be installed"); let subs = h.subs_of(Lang::Java, "ILogger"); assert!( subs.iter().any(|s| s == "ConsoleLogger"), "ConsoleLogger missing from ILogger sub-types: {subs:?}" ); assert!( subs.iter().any(|s| s == "FileLogger"), "FileLogger missing from ILogger sub-types: {subs:?}" ); // Runtime widening: receiver typed as ILogger must reach every // concrete impl's `log(s)`. let widened = gs.resolve_callee_widened(&CalleeQuery { name: "log", caller_lang: Lang::Java, caller_namespace: "src/Main.java", caller_container: None, receiver_type: Some("ILogger"), namespace_qualifier: None, receiver_var: None, arity: Some(1), }); let containers: Vec<&str> = widened.iter().map(|k| k.container.as_str()).collect(); assert!( containers.contains(&"ConsoleLogger"), "ConsoleLogger::log missing from widened set: {containers:?}" ); assert!( containers.contains(&"FileLogger"), "FileLogger::log missing from widened set: {containers:?}" ); } // ───────────────────────────────────────────────────────────────────────── // C2, Rust trait fan-out // ───────────────────────────────────────────────────────────────────────── /// Pass-1 must extract `impl Logger for SafeLogger` and /// `impl Logger for EvalLogger` edges. Receiver typed as `Logger` /// widens to both impls. #[test] fn rust_trait_with_two_impls_fans_out() { let trait_def = br#" pub trait Logger { fn log(&self, s: &str); } "#; let safe_impl = br#" use crate::Logger; pub struct SafeLogger; impl Logger for SafeLogger { fn log(&self, _s: &str) { // no-op } } "#; let eval_impl = br#" use crate::Logger; use std::process::Command; pub struct EvalLogger; impl Logger for EvalLogger { fn log(&self, s: &str) { let _ = Command::new(s).output(); } } "#; let gs = build_gs(&[ File { namespace: "src/lib.rs", bytes: trait_def, }, File { namespace: "src/safe.rs", bytes: safe_impl, }, File { namespace: "src/eval.rs", bytes: eval_impl, }, ]); let h = gs.hierarchy().expect("hierarchy must be installed"); let subs = h.subs_of(Lang::Rust, "Logger"); assert!( subs.iter().any(|s| s == "SafeLogger"), "SafeLogger missing from Logger impls: {subs:?}" ); assert!( subs.iter().any(|s| s == "EvalLogger"), "EvalLogger missing from Logger impls: {subs:?}" ); let widened = gs.resolve_callee_widened(&CalleeQuery { name: "log", caller_lang: Lang::Rust, caller_namespace: "src/main.rs", caller_container: None, receiver_type: Some("Logger"), namespace_qualifier: None, receiver_var: None, arity: Some(2), }); // `arity = 2` because the trait method takes `(&self, &str)`. // Some Rust pipelines record the receiver in arity, others don't , // accept either as long as both impls fan out. let widened_any_arity = if widened.is_empty() { gs.resolve_callee_widened(&CalleeQuery { name: "log", caller_lang: Lang::Rust, caller_namespace: "src/main.rs", caller_container: None, receiver_type: Some("Logger"), namespace_qualifier: None, receiver_var: None, arity: Some(1), }) } else { widened }; let containers: Vec<&str> = widened_any_arity .iter() .map(|k| k.container.as_str()) .collect(); assert!( containers.contains(&"SafeLogger") || containers.contains(&"EvalLogger"), "neither SafeLogger nor EvalLogger present in widened set: {containers:?}; \ hierarchy_edges from impl Logger for X must reach \ resolve_callee_widened" ); } // ───────────────────────────────────────────────────────────────────────── // C3, TypeScript class extends fan-out // ───────────────────────────────────────────────────────────────────────── /// Pass-1 must extract `class Sub extends Super` and /// `class Sub2 extends Super` edges. Receiver typed as `Super` /// widens to both subs. #[test] fn ts_class_with_two_subclasses_fans_out() { let super_class = br#" export class Base { handle(s: string): void {} } "#; let sub_a = br#" import { Base } from './base'; export class SubA extends Base { handle(s: string): void { eval(s); } } "#; let sub_b = br#" import { Base } from './base'; export class SubB extends Base { handle(s: string): void { // safe } } "#; let gs = build_gs(&[ File { namespace: "src/base.ts", bytes: super_class, }, File { namespace: "src/suba.ts", bytes: sub_a, }, File { namespace: "src/subb.ts", bytes: sub_b, }, ]); let h = gs.hierarchy().expect("hierarchy must be installed"); let subs = h.subs_of(Lang::TypeScript, "Base"); assert!( subs.iter().any(|s| s == "SubA"), "SubA missing from Base sub-types: {subs:?}" ); assert!( subs.iter().any(|s| s == "SubB"), "SubB missing from Base sub-types: {subs:?}" ); } // ───────────────────────────────────────────────────────────────────────── // C4, Python class hierarchy // ───────────────────────────────────────────────────────────────────────── /// Pass-1 must extract `class Concrete(Base)` edges. The /// hierarchy index keyed on Python's `Lang::Python` reflects this. #[test] fn python_class_with_subclass_fans_out() { let base_py = br#" class Base: def run(self, s): pass "#; let concrete_py = br#" from base import Base class Concrete(Base): def run(self, s): eval(s) "#; let gs = build_gs(&[ File { namespace: "src/base.py", bytes: base_py, }, File { namespace: "src/concrete.py", bytes: concrete_py, }, ]); let h = gs.hierarchy().expect("hierarchy must be installed"); let subs = h.subs_of(Lang::Python, "Base"); assert!( subs.iter().any(|s| s == "Concrete"), "Concrete missing from Base sub-types in Python: {subs:?}" ); } // ───────────────────────────────────────────────────────────────────────── // C5, Languages without an extractor are silently empty // ───────────────────────────────────────────────────────────────────────── /// Go's structural / implicit interface satisfaction is intractable /// to enumerate from per-file information and is **deliberately /// omitted** from the extractor. This test pins the contract: a Go /// program with what looks like inheritance produces an empty /// hierarchy index, and `resolve_callee_widened` collapses to today's /// single-result behaviour, no fan-out, no regression. #[test] fn go_program_produces_empty_hierarchy() { // Go interface + struct that satisfies it implicitly. No `extends` // syntax exists in Go; the extractor returns no edges. let go_src = br#" package main type Logger interface { Log(s string) } type ConsoleLogger struct{} func (c *ConsoleLogger) Log(s string) { println(s) } "#; let gs = build_gs(&[File { namespace: "src/main.go", bytes: go_src, }]); let h = gs .hierarchy() .expect("hierarchy must be installed even when empty"); assert!( h.subs_of(Lang::Go, "Logger").is_empty(), "Go must have no recorded subtypes — implicit interface satisfaction \ is deliberately omitted" ); // Runtime widening collapses to today's single-result behaviour. let widened = gs.resolve_callee_widened(&CalleeQuery { name: "Log", caller_lang: Lang::Go, caller_namespace: "src/main.go", caller_container: None, receiver_type: Some("Logger"), namespace_qualifier: None, receiver_var: None, arity: Some(1), }); // Either empty (Logger has no Log method body in summaries) or // single result, must NEVER fan out. assert!( widened.len() <= 1, "Go must produce ≤ 1 result with no hierarchy fan-out, got {widened:?}" ); } // ───────────────────────────────────────────────────────────────────────── // C6, Hierarchy install is idempotent // ───────────────────────────────────────────────────────────────────────── /// Calling `install_hierarchy` twice produces the same view. This /// guards against a future regression where a stateful builder leaks /// state across calls. #[test] fn install_hierarchy_is_idempotent() { let logger_iface = br#" package app; public interface ILogger { void log(String s); } "#; let console_logger = br#" package app; public class ConsoleLogger implements ILogger { public void log(String s) { System.out.println(s); } } "#; let mut gs = build_gs(&[ File { namespace: "src/ILogger.java", bytes: logger_iface, }, File { namespace: "src/ConsoleLogger.java", bytes: console_logger, }, ]); let first = gs .hierarchy() .unwrap() .subs_of(Lang::Java, "ILogger") .to_vec(); gs.install_hierarchy(); let second = gs .hierarchy() .unwrap() .subs_of(Lang::Java, "ILogger") .to_vec(); assert_eq!(first, second, "install_hierarchy must be idempotent"); }