expr: anchor every top-level alternative of a regex

src/uu/expr/src/syntax_tree.rs

@@ -275,11 +275,6 @@
     let mut is_start_of_expression = true;
     let mut after_anchor_caret = false;
 
-    // All patterns are anchored so they begin with a caret (^)
-    if pattern_chars.peek() != Some(&'^') {
-        re_string.push('^');
-    }
-
     while let Some(curr) = pattern_chars.next() {
         let curr_is_escaped = prev == '\\' && !prev_is_escaped;
         let mut next_is_start = false;
@@ -425,7 +420,9 @@
 
     let re_string = transpile_bre_to_ere(&pattern_str)?;
 
-    RegexBuilder::new(&format!("(?s){re_string}"))
+    // Patterns are anchored at the start of the string. The group makes the
+    // anchor apply to every top-level `\|` alternative, not just the first one.
+    RegexBuilder::new(&format!("(?s)^(?:{re_string})"))
         .oniguruma_mode(true)
         .leftmost_longest(true)
         .build()
@@ -1504,6 +1501,27 @@
 
         let result = evaluate_match_expression(b"world".to_vec(), b"world$".to_vec()).unwrap();
         assert_eq!(result.eval_as_string(), b"5");
+
+        // Every top-level alternative is anchored, not just the first one
+        let result = evaluate_match_expression(b"xab".to_vec(), b"a\\|b".to_vec()).unwrap();
+        assert_eq!(result.eval_as_string(), b"0");
+
+        let result = evaluate_match_expression(b"xab".to_vec(), b"^a\\|b".to_vec()).unwrap();
+        assert_eq!(result.eval_as_string(), b"0");
+
+        let result = evaluate_match_expression(
+            b"evil status".to_vec(),
+            b"start$\\|stop$\\|status$".to_vec(),
+        )
+        .unwrap();
+        assert_eq!(result.eval_as_string(), b"0");
+
+        let result =
+            evaluate_match_expression(b"zabc".to_vec(), b"x\\|\\(abc\\)".to_vec()).unwrap();
+        assert_eq!(result.eval_as_string(), b"");
+
+        let result = evaluate_match_expression(b"bx".to_vec(), b"a\\|b".to_vec()).unwrap();
+        assert_eq!(result.eval_as_string(), b"1");
     }
 
     #[test]

tests/by-util/test_expr.rs

@@ -288,6 +288,38 @@ fn test_regex_trailing_backslash() {
         .stderr_only("expr: Trailing backslash\n");
 }
 
+#[test]
+fn test_regex_alternation_is_anchored() {
+    new_ucmd!()
+        .args(&["xab", ":", "a\\|b"])
+        .fails_with_code(1)
+        .stdout_only("0\n");
+    new_ucmd!()
+        .args(&["xab", ":", "^a\\|b"])
+        .fails_with_code(1)
+        .stdout_only("0\n");
+    new_ucmd!()
+        .args(&["file:///etc/shadow#ssh://", ":", "https://\\|ssh://"])
+        .fails_with_code(1)
+        .stdout_only("0\n");
+    new_ucmd!()
+        .args(&["evil status", ":", "start$\\|stop$\\|status$"])
+        .fails_with_code(1)
+        .stdout_only("0\n");
+    new_ucmd!()
+        .args(&["zabc", ":", "x\\|\\(abc\\)"])
+        .fails_with_code(1)
+        .stdout_only("\n");
+    new_ucmd!()
+        .args(&["match", "xab", "a\\|b"])
+        .fails_with_code(1)
+        .stdout_only("0\n");
+    new_ucmd!()
+        .args(&["ssh://host", ":", "https://\\|ssh://"])
+        .succeeds()
+        .stdout_only("6\n");
+}
+
 #[test]
 fn test_regex_caret() {
     new_ucmd!()