From d997b53c8d4bf9589a8315f5bb422d2f0d6b216f Mon Sep 17 00:00:00 2001
From: Ethiraric <ethiraric@gmail.com>
Date: Sat, 19 Oct 2024 18:53:51 +0200
Subject: [PATCH] Add fuzz test and fix it.

---
 parser/src/scanner.rs |  4 ++--
 parser/tests/fuzz.rs  | 42 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 44 insertions(+), 2 deletions(-)
 create mode 100644 parser/tests/fuzz.rs

diff --git a/parser/src/scanner.rs b/parser/src/scanner.rs
index 0509181..e97361e 100644
--- a/parser/src/scanner.rs
+++ b/parser/src/scanner.rs
@@ -2249,7 +2249,7 @@ impl<T: Input> Scanner<T> {
             }
 
             // Process blank characters.
-            self.input.lookahead(1);
+            self.input.lookahead(2);
             while self.input.next_is_blank_or_break() {
                 if self.input.next_is_blank() {
                     if !self.leading_whitespace {
@@ -2280,7 +2280,7 @@ impl<T: Input> Scanner<T> {
                         self.leading_whitespace = true;
                     }
                 }
-                self.input.lookahead(1);
+                self.input.lookahead(2);
             }
 
             // check indentation level
diff --git a/parser/tests/fuzz.rs b/parser/tests/fuzz.rs
new file mode 100644
index 0000000..3076c8c
--- /dev/null
+++ b/parser/tests/fuzz.rs
@@ -0,0 +1,42 @@
+use core::str;
+
+use saphyr_parser::{Event, Parser, ScanError};
+
+/// Run the parser through the string.
+///
+/// The parser is run through both the `StrInput` and `BufferedInput` variants. The resulting
+/// events are then compared and must match.
+///
+/// # Returns
+/// This function returns the events if parsing succeeds, the error the parser returned otherwise.
+///
+/// # Panics
+/// This function panics if there is a mismatch between the 2 parser invocations with the different
+/// input traits.
+fn run_parser(input: &str) -> Result<Vec<Event>, ScanError> {
+    let mut str_events = vec![];
+    let mut iter_events = vec![];
+
+    for x in Parser::new_from_str(input) {
+        str_events.push(x?.0);
+    }
+    for x in Parser::new_from_iter(input.chars()) {
+        iter_events.push(x?.0);
+    }
+
+    assert_eq!(str_events, iter_events);
+
+    Ok(str_events)
+}
+
+#[test]
+fn fuzz_1() {
+    // Crashing with an index out-of-bounds error.
+    // In `scan_plain_scalar`, we would lookahead 1 and call `skip_break`, which requires a
+    // lookahead of 2.
+    let raw_input: &[u8] = &[
+        1, 39, 110, 117, 108, 108, 34, 13, 13, 13, 13, 13, 10, 13, 13, 13, 13,
+    ];
+    let s = str::from_utf8(raw_input).unwrap();
+    let _ = run_parser(s);
+}