Merge pull request #7997 from frendsick/fix/expr-regex-range-quantifiers

expr: Fix parsing range quantifiers in regex

src/uu/expr/src/syntax_tree.rs

@@ -160,7 +160,6 @@ impl StringOp {
                 let first = pattern_chars.next();
                 match first {
                     Some('^') => {} // Start of string anchor is already added
-                    Some('*') => re_string.push_str(r"\*"),
                     Some('$') if !is_end_of_expression(&pattern_chars) => re_string.push_str(r"\$"),
                     Some('\\') if right.len() == 1 => return Err(ExprError::TrailingBackslash),
                     Some(char) => re_string.push(char),
@@ -191,10 +190,17 @@ impl StringOp {
                         '\\' if !curr_is_escaped && pattern_chars.peek().is_none() => {
                             return Err(ExprError::TrailingBackslash);
                         }
+                        '{' if curr_is_escaped && is_valid_range_quantifier(&pattern_chars) => {
+                            re_string.push(curr);
+                            // Set the lower bound of range quantifier to 0 if it is missing
+                            if pattern_chars.peek() == Some(&',') {
+                                re_string.push('0');
+                            }
+                        }
                         _ => re_string.push(curr),
                     }
 
-                    prev_is_escaped = prev == '\\' && !prev_is_escaped;
+                    prev_is_escaped = curr_is_escaped;
                     prev = curr;
                 }
 
@@ -244,6 +250,46 @@ where
     }
 }
 
+/// Check if regex pattern character iterator is at the start of a valid range quantifier.
+/// The iterator's start position is expected to be after the opening brace.
+/// Range quantifier ends to closing brace.
+///
+/// # Examples of valid range quantifiers
+///
+/// - `r"\{3\}"`
+/// - `r"\{3,\}"`
+/// - `r"\{,6\}"`
+/// - `r"\{3,6\}"`
+/// - `r"\{,\}"`
+fn is_valid_range_quantifier<I>(pattern_chars: &I) -> bool
+where
+    I: Iterator<Item = char> + Clone,
+{
+    // Parse the string between braces
+    let mut quantifier = String::new();
+    let mut pattern_chars_clone = pattern_chars.clone().peekable();
+    let Some(mut prev) = pattern_chars_clone.next() else {
+        return false;
+    };
+    let mut prev_is_escaped = false;
+    while let Some(curr) = pattern_chars_clone.next() {
+        if prev == '\\' && curr == '}' && !prev_is_escaped {
+            break;
+        }
+        if pattern_chars_clone.peek().is_none() {
+            return false;
+        }
+
+        quantifier.push(prev);
+        prev_is_escaped = prev == '\\' && !prev_is_escaped;
+        prev = curr;
+    }
+
+    // Check if parsed quantifier is valid
+    let re = Regex::new(r"(\d+|\d*,\d*)").expect("valid regular expression");
+    re.is_match(&quantifier)
+}
+
 /// Check for errors in a supplied regular expression
 ///
 /// GNU coreutils shows messages for invalid regular expressions
@@ -287,10 +333,7 @@ fn check_posix_regex_errors(pattern: &str) -> ExprResult<()> {
                         .expect("splitn always returns at least one string"),
                     repetition.next(),
                 ) {
-                    ("", None) => {
+                    ("", Some("")) => {}
-                        // Empty repeating pattern
-                        invalid_content_error = true;
-                    }
                     (x, None | Some("")) => {
                         if x.parse::<i16>().is_err() {
                             invalid_content_error = true;
@@ -750,6 +793,7 @@ pub fn is_truthy(s: &NumOrStr) -> bool {
 mod test {
     use crate::ExprError;
     use crate::ExprError::InvalidBracketContent;
+    use crate::syntax_tree::is_valid_range_quantifier;
 
     use super::{
         AstNode, AstNodeInner, BinOp, NumericOp, RelationOp, StringOp, check_posix_regex_errors,
@@ -998,4 +1042,22 @@ mod test {
             Err(InvalidBracketContent)
         );
     }
+
+    #[test]
+    fn test_is_valid_range_quantifier() {
+        assert!(is_valid_range_quantifier(&"3\\}".chars()));
+        assert!(is_valid_range_quantifier(&"3,\\}".chars()));
+        assert!(is_valid_range_quantifier(&",6\\}".chars()));
+        assert!(is_valid_range_quantifier(&"3,6\\}".chars()));
+        assert!(is_valid_range_quantifier(&",\\}".chars()));
+        assert!(is_valid_range_quantifier(&"3,6\\}anything".chars()));
+        assert!(!is_valid_range_quantifier(&"\\{3,6\\}".chars()));
+        assert!(!is_valid_range_quantifier(&"\\}".chars()));
+        assert!(!is_valid_range_quantifier(&"".chars()));
+        assert!(!is_valid_range_quantifier(&"3".chars()));
+        assert!(!is_valid_range_quantifier(&"3,".chars()));
+        assert!(!is_valid_range_quantifier(&",6".chars()));
+        assert!(!is_valid_range_quantifier(&"3,6".chars()));
+        assert!(!is_valid_range_quantifier(&",".chars()));
+    }
 }

tests/by-util/test_expr.rs

@@ -875,7 +875,6 @@ mod gnu_expr {
             .stdout_only("\n");
     }
 
-    #[ignore]
     #[test]
     fn test_bre17() {
         new_ucmd!()
@@ -884,7 +883,6 @@ mod gnu_expr {
             .stdout_only("{1}a\n");
     }
 
-    #[ignore]
     #[test]
     fn test_bre18() {
         new_ucmd!()
@@ -893,7 +891,6 @@ mod gnu_expr {
             .stdout_only("1\n");
     }
 
-    #[ignore]
     #[test]
     fn test_bre19() {
         new_ucmd!()
@@ -1105,7 +1102,6 @@ mod gnu_expr {
             .stderr_contains("Invalid content of \\{\\}");
     }
 
-    #[ignore]
     #[test]
     fn test_bre45() {
         new_ucmd!()
@@ -1114,7 +1110,6 @@ mod gnu_expr {
             .stdout_only("1\n");
     }
 
-    #[ignore]
     #[test]
     fn test_bre46() {
         new_ucmd!()