Avoid infinite loops in many0! and many1!

Fix for #27

Author: Geal

CHANGELOG.md

@@ -3,6 +3,8 @@
 ## [Unreleased][unreleased]
 
 ### Changed
+- `many0!` and `many1!` forbid parsers that do not consume input
+- `is_a!`, `is_not!`, `alpha`, `digit`, `space`, `multispace` will now return an error if they do not consume at least one byte
 
 ## 0.2.1 - 2015-04-04
 

src/macros.rs

@@ -433,7 +433,11 @@ macro_rules! is_not(
         }
         if parsed { break; }
       }
-      IResult::Done(&$input[index..], &$input[0..index])
+      if index == 0 {
+        IResult::Error(0)
+      } else {
+        IResult::Done(&$input[index..], &$input[0..index])
+      }
     }
   );
 );
@@ -473,7 +477,11 @@ macro_rules! is_a(
         }
         if !cont { break; }
       }
-      IResult::Done(&$input[index..], &$input[0..index])
+      if index == 0 {
+        IResult::Error(0)
+      } else {
+        IResult::Done(&$input[index..], &$input[0..index])
+      }
     }
   );
 );
@@ -500,7 +508,11 @@ macro_rules! filter(
           break;
         }
       }
-      IResult::Done(&$input[index..], &$input[0..index])
+      if index == 0 {
+        IResult::Error(0)
+      } else {
+        IResult::Done(&$input[index..], &$input[0..index])
+      }
     }
   );
 );
@@ -605,6 +617,9 @@ macro_rules! many0(
       loop {
         match $submac!(&$i[begin..], $($args)*) {
           IResult::Done(i,o) => {
+            if i.len() == $i[begin..].len() {
+              break;
+            }
             res.push(o);
             begin += remaining - i.len();
             remaining = i.len();
@@ -622,7 +637,7 @@ macro_rules! many0(
   );
 );
 
-/// Applies the parser 0 or more times and returns the list of results in a Vec
+/// Applies the parser 1 or more times and returns the list of results in a Vec
 ///
 /// the embedded parser may return Incomplete
 ///
@@ -646,6 +661,9 @@ macro_rules! many1(
       loop {
         match $submac!(&$i[begin..], $($args)*) {
           IResult::Done(i,o) => {
+            if i.len() == $i[begin..].len() {
+              break;
+            }
             res.push(o);
             begin += remaining - i.len();
             remaining = i.len();
@@ -655,7 +673,7 @@ macro_rules! many1(
           }
         }
       }
-      if begin == 0 {
+      if res.len() == 0 {
         IResult::Error(0)
       } else {
         IResult::Done(&$i[begin..], res)
@@ -1012,7 +1030,7 @@ mod tests {
     assert_eq!(a_or_b(b), Done(&b"cde"[..], &b"b"[..]));
 
     let c = &b"cdef"[..];
-    assert_eq!(a_or_b(c), Done(&b"cdef"[..], &b""[..]));
+    assert_eq!(a_or_b(c), Error(0));
 
     let d = &b"bacdef"[..];
     assert_eq!(a_or_b(d), Done(&b"cdef"[..], &b"ba"[..]));
@@ -1186,6 +1204,22 @@ mod tests {
     assert_eq!(multi(c), Error(0));
   }
 
+  #[test]
+  fn infinite_many() {
+    fn tst(input: &[u8]) -> IResult<&[u8], &[u8]> {
+      Error(0)
+    }
+
+    // should not go into an infinite loop
+    named!(multi0<&[u8],Vec<&[u8]> >, many0!(tst));
+    let a = &b"abcdef"[..];
+    assert_eq!(multi0(a), Done(a, Vec::new()));
+
+    named!(multi1<&[u8],Vec<&[u8]> >, many1!(tst));
+    let a = &b"abcdef"[..];
+    assert_eq!(multi1(a), Error(0));
+  }
+
   #[test]
   fn take_until_test() {
     named!(x, take_until_and_consume!("efgh"));

src/nom.rs

@@ -75,7 +75,11 @@ pub fn is_space(chr:u8) -> bool {
 pub fn alpha(input:&[u8]) -> IResult<&[u8], &[u8]> {
   for idx in 0..input.len() {
     if !is_alphabetic(input[idx]) {
-      return Done(&input[idx..], &input[0..idx])
+      if idx == 0 {
+        return Error(0)
+      } else {
+        return Done(&input[idx..], &input[0..idx])
+      }
     }
   }
   Done(b"", input)
@@ -84,7 +88,11 @@ pub fn alpha(input:&[u8]) -> IResult<&[u8], &[u8]> {
 pub fn digit(input:&[u8]) -> IResult<&[u8], &[u8]> {
   for idx in 0..input.len() {
     if !is_digit(input[idx]) {
-      return Done(&input[idx..], &input[0..idx])
+      if idx == 0 {
+        return Error(0)
+      } else {
+        return Done(&input[idx..], &input[0..idx])
+      }
     }
   }
   Done(b"", input)
@@ -93,7 +101,11 @@ pub fn digit(input:&[u8]) -> IResult<&[u8], &[u8]> {
 pub fn alphanumeric(input:&[u8]) -> IResult<&[u8], &[u8]> {
   for idx in 0..input.len() {
     if !is_alphanumeric(input[idx]) {
-      return Done(&input[idx..], &input[0..idx])
+      if idx == 0 {
+        return Error(0)
+      } else {
+        return Done(&input[idx..], &input[0..idx])
+      }
     }
   }
   Done(b"", input)
@@ -102,16 +114,25 @@ pub fn alphanumeric(input:&[u8]) -> IResult<&[u8], &[u8]> {
 pub fn space(input:&[u8]) -> IResult<&[u8], &[u8]> {
   for idx in 0..input.len() {
     if !is_space(input[idx]) {
-      return Done(&input[idx..], &input[0..idx])
+      if idx == 0 {
+        return Error(0)
+      } else {
+        return Done(&input[idx..], &input[0..idx])
+      }
     }
   }
   Done(b"", input)
 }
 
 pub fn multispace(input:&[u8]) -> IResult<&[u8], &[u8]> {
   for idx in 0..input.len() {
+    println!("multispace at index: {}", idx);
     if !is_space(input[idx]) && input[idx] != '\r' as u8 && input[idx] != '\n' as u8 {
-      return Done(&input[idx..], &input[0..idx])
+      if idx == 0 {
+        return Error(0)
+      } else {
+        return Done(&input[idx..], &input[0..idx])
+      }
     }
   }
   Done(b"", input)
@@ -267,13 +288,13 @@ mod tests {
     let d: &[u8] = "azé12".as_bytes();
     let e: &[u8] = b" ";
     assert_eq!(alpha(a), Done(empty, a));
-    assert_eq!(alpha(b), Done(b, empty));
+    assert_eq!(alpha(b), Error(0));
     assert_eq!(alpha(c), Done(&c[1..], &b"a"[..]));
     assert_eq!(alpha(d), Done("é12".as_bytes(), &b"az"[..]));
-    assert_eq!(digit(a), Done(a, empty));
+    assert_eq!(digit(a), Error(0));
     assert_eq!(digit(b), Done(empty, b));
-    assert_eq!(digit(c), Done(c, empty));
-    assert_eq!(digit(d), Done(d, empty));
+    assert_eq!(digit(c), Error(0));
+    assert_eq!(digit(d), Error(0));
     assert_eq!(alphanumeric(a), Done(empty, a));
     assert_eq!(alphanumeric(b), Done(empty, b));
     assert_eq!(alphanumeric(c), Done(empty, c));