From 80c66dd29c0ed0d4188b1e144a631432bb9b1a57 Mon Sep 17 00:00:00 2001
From: Tristan F <26509014+LeoDog896@users.noreply.github.com>
Date: Thu, 7 Mar 2024 18:52:20 -0500
Subject: [PATCH] feat: base safe test suite

---
 crates/redos/src/ilq.rs           |  5 ++--
 crates/redos/src/ir.rs            |  2 +-
 crates/redos/src/lib.rs           |  2 +-
 crates/redos/src/vulnerability.rs | 13 ++++++++--
 crates/redos/tests/lib.rs         | 41 ++++++++++++++++++++++++++-----
 crates/redos/tests/safe.txt       | 30 ++++++++++++++++++++++
 fixtures/redos.toml               |  1 -
 7 files changed, 81 insertions(+), 13 deletions(-)
 create mode 100644 crates/redos/tests/safe.txt
 delete mode 100644 fixtures/redos.toml

diff --git a/crates/redos/src/ilq.rs b/crates/redos/src/ilq.rs
index 041ed96..18d9b86 100644
--- a/crates/redos/src/ilq.rs
+++ b/crates/redos/src/ilq.rs
@@ -46,10 +46,11 @@ pub fn scan_ilq(expr: &Expr) -> IlqReturn {
         // luckily, we can just pretend as if the child is the root of its own tree
         Expr::Repeat(e) => scan_ilq(e),
 
+        // TODO: proper support for lookarounds
         Expr::LookAround(e, _) => scan_ilq(e),
 
-        // TODO: atomic groups and lookarounds
-        _ => IlqReturn::new(true),
+        // TODO: proper support for atomic groups
+        Expr::AtomicGroup(e) => scan_ilq(e),
     }
 }
 
diff --git a/crates/redos/src/ir.rs b/crates/redos/src/ir.rs
index ebc538e..53779a8 100644
--- a/crates/redos/src/ir.rs
+++ b/crates/redos/src/ir.rs
@@ -123,7 +123,7 @@ pub fn to_expr(
             let range = hi - lo;
 
             let expression = to_expr(child, config, group_increment);
-            let expression = if range > config.max_quantifier {
+            let expression = if range > config.four_max_quantifier {
                 expression.map(|child| Expr::Repeat(Box::new(child)))
             } else {
                 expression
diff --git a/crates/redos/src/lib.rs b/crates/redos/src/lib.rs
index 82f5cc1..2e425d3 100644
--- a/crates/redos/src/lib.rs
+++ b/crates/redos/src/lib.rs
@@ -34,7 +34,7 @@ impl RegexInfo {
 ///
 /// A regex must meet the following criteria to be even considered to be vulnerable:
 /// - It must contain a repeat
-/// - The repeat must have a bound size greater than `config.max_quantifier`
+/// - The repeat must have a bound size greater than `config.second_max_quantifier`
 /// - The regex must have a terminating state (to allow for backtracking) (TODO: this is not implemented yet)
 fn regex_pre_scan(expr: &Expr) -> RegexInfo {
     match expr {
diff --git a/crates/redos/src/vulnerability.rs b/crates/redos/src/vulnerability.rs
index 7900f52..6b504e2 100644
--- a/crates/redos/src/vulnerability.rs
+++ b/crates/redos/src/vulnerability.rs
@@ -9,7 +9,9 @@ pub enum Vulnerability {
 impl Default for VulnerabilityConfig {
     fn default() -> Self {
         Self {
-            max_quantifier: 100,
+            two_max_quantifier: 1000,
+            three_max_quantifier: 100,
+            four_max_quantifier: 10,
         }
     }
 }
@@ -23,5 +25,12 @@ pub enum Complexity {
 #[derive(Debug, PartialEq, Eq, Clone, Copy)]
 pub struct VulnerabilityConfig {
     /// The max size an upper bound can be before it is considered "large"
-    pub max_quantifier: usize,
+    /// in a superlinear fashion
+    pub two_max_quantifier: usize,
+    /// The max size an upper bound can be before it is considered "large"
+    /// for 3rd degree polynomials
+    pub three_max_quantifier: usize,
+    /// The max size an upper bound can be before it is considered "large"
+    /// for 4th degree polynomials
+    pub four_max_quantifier: usize,
 }
diff --git a/crates/redos/tests/lib.rs b/crates/redos/tests/lib.rs
index ebd8a62..88702e0 100644
--- a/crates/redos/tests/lib.rs
+++ b/crates/redos/tests/lib.rs
@@ -2,19 +2,48 @@
 mod tests {
     use redos::vulnerabilities;
 
-    fn assert_safe(regex: &str) {
+    static SAFE: &str = include_str!("safe.txt");
+
+    /// Takes a test file and returns pairs
+    /// of test names and their contents
+    fn parse_test_file(file: &str) -> Vec<(String, Vec<String>)> {
+        let mut tests: Vec<(String, Vec<String>)> = vec![];
+
+        for line in file.lines() {
+            if line.starts_with('#') {
+                continue;
+            }
+
+            if line.starts_with('\t') {
+                let mut line = line.to_string();
+                line.pop();
+                tests.last_mut().unwrap().1.push(line);
+            } else {
+                tests.push((line.to_string(), vec![]));
+            }
+        }
+
+        tests
+    }
+
+    fn assert_safe(regex: &str, message: &str) {
         assert_eq!(
             vulnerabilities(regex, &Default::default())
                 .unwrap()
                 .vulnerabilities,
-            vec![]
+            vec![],
+            "{} failed: {}",
+            message,
+            regex
         );
     }
 
     #[test]
-    fn trivial_regexes() {
-        assert_safe("abc");
-        assert_safe("(abc|def)|[nhi]?");
-        assert_safe("a{1,43}");
+    fn check_safe() {
+        for (name, tests) in parse_test_file(SAFE) {
+            for test in tests {
+                assert_safe(&test, &name);
+            }
+        }
     }
 }
diff --git a/crates/redos/tests/safe.txt b/crates/redos/tests/safe.txt
new file mode 100644
index 0000000..d139af8
--- /dev/null
+++ b/crates/redos/tests/safe.txt
@@ -0,0 +1,30 @@
+base example
+        abc
+lets try out groups
+        (abc|def)|[nhi]?
+tiny repeating string
+        a{1,10}
+bigger repeated string that's just right
+        a{}
+nested group
+        (((a)))
+try putting some quantifiers outside
+        ((a+)+)+
+are optionals detected?
+        (a?)+
+lets try lookarounds
+        (?<=a)
+        (?<!a)
+        (?!a)
+        (?=a)
+        (?>a)
+        (?>a+)
+can we decompose alternations?
+        (a+)|(b+)|(((a)))
+how about alternations in lookarounds?
+        (?<=a|b)
+        (?<!a|b)
+        (?!a|b)
+        (?=a|b)
+        (?>a|b)
+        (?>a+|b+)
diff --git a/fixtures/redos.toml b/fixtures/redos.toml
deleted file mode 100644
index 0637a08..0000000
--- a/fixtures/redos.toml
+++ /dev/null
@@ -1 +0,0 @@
-[]
\ No newline at end of file