fix: bug with consecutive repeats.

VirusTotal · Oct 7, 2024 · 85382b4 · 85382b4
1 parent 3d35671
commit 85382b4
Show file tree

Hide file tree

Showing 2 changed files with 27 additions and 16 deletions.
diff --git a/lib/src/re/thompson/pikevm.rs b/lib/src/re/thompson/pikevm.rs
@@ -343,7 +343,7 @@ pub(crate) fn epsilon_closure(
 
     let is_word_char = |c: u8| c == b'_' || c.is_ascii_alphanumeric();
 
-    while let Some(mut ts) = state.threads.pop() {
+    while let Some(ts) = state.threads.pop() {
         let (instr, instr_size) =
             InstrParser::decode_instr(unsafe { code.get_unchecked(ts.ip..) });
         match instr {
@@ -382,12 +382,15 @@ pub(crate) fn epsilon_closure(
                 state.threads.push(ts.ip_offset(instr_size.into()));
             }
             Instr::RepeatGreedyEnd { offset, min, max } => {
-                ts.rep_count += 1;
                 if ts.rep_count >= min {
-                    state.threads.push(ts.ip_offset(instr_size.into()));
+                    let mut ts = ts.ip_offset(instr_size.into());
+                    ts.rep_count = 0;
+                    state.threads.push(ts);
                 }
                 if ts.rep_count < max {
-                    state.threads.push(ts.ip_offset(offset));
+                    let mut ts = ts.ip_offset(offset);
+                    ts.rep_count += 1;
+                    state.threads.push(ts);
                 }
             }
             Instr::RepeatNonGreedyStart { offset, min } => {
@@ -397,12 +400,15 @@ pub(crate) fn epsilon_closure(
                 }
             }
             Instr::RepeatNonGreedyEnd { offset, min, max } => {
-                ts.rep_count += 1;
                 if ts.rep_count < max {
-                    state.threads.push(ts.ip_offset(offset));
+                    let mut ts = ts.ip_offset(offset);
+                    ts.rep_count += 1;
+                    state.threads.push(ts);
                 }
                 if ts.rep_count >= min {
-                    state.threads.push(ts.ip_offset(instr_size.into()));
+                    let mut ts = ts.ip_offset(instr_size.into());
+                    ts.rep_count = 0;
+                    state.threads.push(ts);
                 }
             }
             Instr::Jump(offset) => {

diff --git a/lib/src/tests/mod.rs b/lib/src/tests/mod.rs
@@ -1254,7 +1254,7 @@ fn regexp_patterns_2() {
 }
 
 #[test]
-fn regexp_patterns_2a() {
+fn regexp_patterns_3() {
     pattern_match!(r#"/.b{15}/"#, b"abbbbbbbbbbbbbbb", b"abbbbbbbbbbbbbbb");
     pattern_match!(
         r#"/.b{15,16}/"#,
@@ -1286,6 +1286,16 @@ fn regexp_patterns_2a() {
         b"aabbbbbbbbbbbbbbbcccc",
         b"aabbbbbbbbbbbbbbbcccc"
     );
+    pattern_match!(
+        r#"/abcd.{0,11}efgh.{0,11}ijk/"#,
+        b"abcd123456789ABefgh123456789ABijk",
+        b"abcd123456789ABefgh123456789ABijk"
+    );
+    pattern_match!(
+        r#"/abcd.{0,11}?efgh.{0,11}?ijk/"#,
+        b"abcd123456789ABefgh123456789ABijk",
+        b"abcd123456789ABefgh123456789ABijk"
+    );
     pattern_match!(r#"/ab{2,15}c/"#, b"abbbc", b"abbbc");
     pattern_match!(r#"/ab{2,15}?c/"#, b"abbbc", b"abbbc");
     pattern_match!(r#"/ab{0,15}?c/"#, b"abc", b"abc");
@@ -1348,7 +1358,7 @@ fn regexp_patterns_2a() {
 }
 
 #[test]
-fn regexp_patterns_3() {
+fn regexp_patterns_4() {
     pattern_match!(r#"/a[bx]c/"#, b"abc", b"abc");
     pattern_match!(r#"/a[bx]c/"#, b"axc", b"axc");
     pattern_match!(r#"/a[0-9]*b/"#, b"ab", b"ab");
@@ -1469,7 +1479,7 @@ fn regexp_patterns_3() {
 }
 
 #[test]
-fn regexp_patterns_4() {
+fn regexp_patterns_5() {
     pattern_match!(r"/\\/", b"\\", b"\\");
     pattern_match!(r"/\babc/", b"abc", b"abc");
     pattern_match!(r"/abc\b/", b"abc", b"abc");
@@ -1539,7 +1549,7 @@ fn regexp_patterns_4() {
 }
 
 #[test]
-fn regexp_patterns_5() {
+fn regexp_patterns_6() {
     rule_true!(
         r#"rule test {
             strings:
@@ -3428,8 +3438,3 @@ fn test_defined_3() {
     condition_true!(r#"not test_proto3.bool_undef"#);
     condition_true!(r#"test_proto3.string_undef == """#);
 }
-
-#[test]
-fn issue() {
-    pattern_match!(r#"/a.(bc.){2}/"#, b"aabcabca", b"aabcabca");
-}