Initial commit: noxis JVM bytecode obfuscator
Workspace of 9 crates (noxis, noxis-core, noxis-classfile, noxis-analysis, noxis-passes, noxis-runtime, noxis-jar, noxis-verify, noxis-structure-check). AGPL-3.0-only license, Rust 2024 edition, unsafe_code denied.
This commit is contained in:
commit
a641e69455
140 changed files with 17197 additions and 0 deletions
215
crates/noxis-passes/src/namegen/mod.rs
Normal file
215
crates/noxis-passes/src/namegen/mod.rs
Normal file
|
|
@ -0,0 +1,215 @@
|
|||
//! Генерация имён: стили, анти-ключевые слова, уникальность в scope.
|
||||
|
||||
pub mod alphabet;
|
||||
|
||||
use std::collections::HashSet;
|
||||
|
||||
use noxis_core::{NameStyle, Rng};
|
||||
|
||||
use self::alphabet::{
|
||||
AVOID_CLASS, AVOID_MEMBER, CONFUSABLE, DECOY_LEFT, DECOY_RIGHT, KEYWORDS, LATIN_LOWER,
|
||||
LATIN_UPPER, UNICODE_LETTERS,
|
||||
};
|
||||
|
||||
/// Генератор имён с гарантированной уникальностью в пределах одного scope.
|
||||
pub struct NameGenerator {
|
||||
rng: Rng,
|
||||
style: NameStyle,
|
||||
used: HashSet<String>,
|
||||
/// Счётчик-страховка, если случайные варианты исчерпаны.
|
||||
fallback: u64,
|
||||
avoid_class: bool,
|
||||
}
|
||||
|
||||
impl NameGenerator {
|
||||
/// Генератор для указанного стиля.
|
||||
pub fn new(rng: Rng, style: NameStyle) -> Self {
|
||||
Self {
|
||||
rng,
|
||||
style,
|
||||
used: HashSet::new(),
|
||||
fallback: 0,
|
||||
avoid_class: false,
|
||||
}
|
||||
}
|
||||
|
||||
/// Режим «имена классов»: дополнительно избегаем имён из JDK/аннотаций.
|
||||
pub fn for_classes(mut self) -> Self {
|
||||
self.avoid_class = true;
|
||||
self
|
||||
}
|
||||
|
||||
/// Занять конкретное имя (защита от коллизий с keep-именами).
|
||||
pub fn reserve(&mut self, name: impl Into<String>) {
|
||||
self.used.insert(name.into());
|
||||
}
|
||||
|
||||
/// Следующее свободное имя.
|
||||
pub fn next_name(&mut self) -> String {
|
||||
for _ in 0..64 {
|
||||
let candidate = self.candidate();
|
||||
if self.acceptable(&candidate) {
|
||||
self.used.insert(candidate.clone());
|
||||
return candidate;
|
||||
}
|
||||
}
|
||||
// Страховка: детерминированный счётчик, гарантированно уникален.
|
||||
loop {
|
||||
self.fallback += 1;
|
||||
let candidate = format!("{}{}", self.style_prefix(), self.fallback);
|
||||
if self.acceptable(&candidate) {
|
||||
self.used.insert(candidate.clone());
|
||||
return candidate;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
fn style_prefix(&self) -> &'static str {
|
||||
match self.style {
|
||||
NameStyle::ShortLatin | NameStyle::IlO0 | NameStyle::MixedLatin => "z",
|
||||
NameStyle::Unicode => "\u{043e}",
|
||||
NameStyle::DecoyWords => "tmp",
|
||||
}
|
||||
}
|
||||
|
||||
fn candidate(&mut self) -> String {
|
||||
match self.style {
|
||||
NameStyle::ShortLatin => {
|
||||
let len = 1 + self.rng.below(3) as usize;
|
||||
random_from(LATIN_LOWER, len, &mut self.rng)
|
||||
}
|
||||
NameStyle::IlO0 => {
|
||||
let len = 3 + self.rng.below(5) as usize;
|
||||
random_from(CONFUSABLE, len, &mut self.rng)
|
||||
}
|
||||
NameStyle::MixedLatin => {
|
||||
let len = 6 + self.rng.below(7) as usize;
|
||||
let mut out = String::with_capacity(len);
|
||||
for i in 0..len {
|
||||
let alpha = if self.rng.below(2) == 0 {
|
||||
LATIN_LOWER
|
||||
} else {
|
||||
LATIN_UPPER
|
||||
};
|
||||
let idx = self.rng.below(alpha.len() as u64) as usize;
|
||||
let ch = alpha[idx] as char;
|
||||
// Первый символ — только буква (не цифра).
|
||||
if i == 0 && !ch.is_ascii_alphabetic() {
|
||||
out.push('a');
|
||||
} else {
|
||||
out.push(ch);
|
||||
}
|
||||
}
|
||||
out
|
||||
}
|
||||
NameStyle::Unicode => {
|
||||
let len = 4 + self.rng.below(6) as usize;
|
||||
(0..len)
|
||||
.map(|_| {
|
||||
let idx = self.rng.below(UNICODE_LETTERS.len() as u64) as usize;
|
||||
UNICODE_LETTERS[idx]
|
||||
})
|
||||
.collect()
|
||||
}
|
||||
NameStyle::DecoyWords => {
|
||||
let l = DECOY_LEFT[self.rng.below(DECOY_LEFT.len() as u64) as usize];
|
||||
let r = DECOY_RIGHT[self.rng.below(DECOY_RIGHT.len() as u64) as usize];
|
||||
let sep = if self.rng.below(2) == 0 { "_" } else { "" };
|
||||
let mut name = format!("{l}{sep}{r}");
|
||||
if self.rng.below(2) == 0 {
|
||||
name = capitalize(&name);
|
||||
}
|
||||
name
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
fn acceptable(&self, name: &str) -> bool {
|
||||
if name.is_empty() || self.used.contains(name) {
|
||||
return false;
|
||||
}
|
||||
if KEYWORDS.contains(&name) {
|
||||
return false;
|
||||
}
|
||||
if self.avoid_class && AVOID_CLASS.contains(&name) {
|
||||
return false;
|
||||
}
|
||||
if !self.avoid_class && AVOID_MEMBER.contains(&name) {
|
||||
return false;
|
||||
}
|
||||
let Some(first) = name.chars().next() else {
|
||||
return false;
|
||||
};
|
||||
if !first.is_alphabetic() && first != '_' && first != '$' {
|
||||
return false;
|
||||
}
|
||||
name.chars()
|
||||
.all(|c| c.is_alphanumeric() || c == '_' || c == '$')
|
||||
}
|
||||
}
|
||||
|
||||
fn random_from(alphabet: &[u8], len: usize, rng: &mut Rng) -> String {
|
||||
(0..len)
|
||||
.map(|_| {
|
||||
let idx = rng.below(alphabet.len() as u64) as usize;
|
||||
alphabet[idx] as char
|
||||
})
|
||||
.collect()
|
||||
}
|
||||
|
||||
fn capitalize(s: &str) -> String {
|
||||
let mut chars = s.chars();
|
||||
match chars.next() {
|
||||
Some(c) => c.to_uppercase().collect::<String>() + chars.as_str(),
|
||||
None => String::new(),
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
#![allow(clippy::expect_used, clippy::unwrap_used)]
|
||||
use super::alphabet::{AVOID_MEMBER, KEYWORDS};
|
||||
use super::*;
|
||||
|
||||
#[test]
|
||||
fn names_are_unique_and_valid() {
|
||||
let mut names = NameGenerator::new(Rng::from_seed(1), NameStyle::ShortLatin);
|
||||
let mut seen = HashSet::new();
|
||||
for _ in 0..500 {
|
||||
let name = names.next_name();
|
||||
assert!(seen.insert(name.clone()), "повтор имени {name}");
|
||||
assert!(!KEYWORDS.contains(&name.as_str()));
|
||||
assert!(name.starts_with(|c: char| c.is_ascii_alphabetic()));
|
||||
}
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn avoids_reserved_member_names() {
|
||||
let mut names = NameGenerator::new(Rng::from_seed(2), NameStyle::DecoyWords);
|
||||
for _ in 0..500 {
|
||||
let name = names.next_name();
|
||||
assert!(!AVOID_MEMBER.contains(&name.as_str()));
|
||||
}
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn unicode_style_produces_identifier_valid_names() {
|
||||
let mut names = NameGenerator::new(Rng::from_seed(3), NameStyle::Unicode);
|
||||
for _ in 0..100 {
|
||||
let name = names.next_name();
|
||||
assert!(
|
||||
name.chars()
|
||||
.all(|c| c.is_alphanumeric() || c == '_' || c == '$')
|
||||
);
|
||||
}
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn reserved_names_never_reissued() {
|
||||
let mut names = NameGenerator::new(Rng::from_seed(4), NameStyle::ShortLatin);
|
||||
names.reserve("abc");
|
||||
for _ in 0..300 {
|
||||
assert_ne!(names.next_name(), "abc");
|
||||
}
|
||||
}
|
||||
}
|
||||
Loading…
Add table
Add a link
Reference in a new issue