Add explicit Unicode mode to all regex patterns for proper Unicode handling

Co-authored-by: imjasonh <210737+imjasonh@users.noreply.github.com>

src/lib.rs

@@ -71,9 +71,9 @@ fn run(params: &mut RunParams, test_data_glob: &str) -> Result<()> {
             (".", test_data_glob)
         };
 
-        // Convert glob pattern to a simple matcher
+        // Convert glob pattern to a simple matcher with Unicode support
         let pattern_regex = pattern.replace("*", ".*");
-        let regex = regex::Regex::new(&format!("^{}$", pattern_regex))?;
+        let regex = regex::Regex::new(&format!("(?u)^{}$", pattern_regex))?;
 
         // Walk the directory and collect matching files
         for entry in WalkDir::new(base_dir).min_depth(1).max_depth(1) {

src/run/commands.rs

@@ -218,7 +218,14 @@ impl TestEnvironment {
 
     /// Search for a pattern in files (like grep)
     pub fn grep_files(&mut self, pattern: &str, files: &[String]) -> Result<()> {
-        let regex = Regex::new(pattern)
+        // Enable Unicode mode for proper Unicode character matching in grep patterns
+        let pattern_with_flags = if pattern.starts_with("(?") {
+            // Pattern already has flags, don't add more
+            pattern.to_string()
+        } else {
+            format!("(?u){}", pattern)
+        };
+        let regex = Regex::new(&pattern_with_flags)
             .map_err(|e| Error::command_error("grep", format!("Invalid regex: {}", e)))?;
 
         let mut _found_matches = false;

src/run/environment.rs

@@ -236,7 +236,9 @@ impl TestEnvironment {
             || expected_substituted.contains('*')
             || expected_substituted.contains('.')
         {
-            let regex_pattern = format!("(?s){}", expected_substituted); // (?s) enables DOTALL mode
+            // Enable DOTALL mode (?s) for . to match newlines
+            // Enable Unicode mode (?u) for proper Unicode character matching
+            let regex_pattern = format!("(?su){}", expected_substituted);
             let regex = Regex::new(&regex_pattern)
                 .map_err(|e| Error::command_error(output_type, format!("Invalid regex: {}", e)))?;
 

tests/unicode_handling.rs

@@ -19,7 +19,7 @@ stdout "Success ✓"
     fs::write(&test_file, test_content).unwrap();
 
     let result = testscript::run(testdata_dir.to_string_lossy()).execute();
-    
+
     assert!(
         result.is_ok(),
         "Unicode stdout exact match should work: {:?}",
@@ -42,7 +42,7 @@ stderr "Error ✗"
     fs::write(&test_file, test_content).unwrap();
 
     let result = testscript::run(testdata_dir.to_string_lossy()).execute();
-    
+
     assert!(
         result.is_ok(),
         "Unicode stderr exact match should work: {:?}",
@@ -65,7 +65,7 @@ stdout "Success .* completed"
     fs::write(&test_file, test_content).unwrap();
 
     let result = testscript::run(testdata_dir.to_string_lossy()).execute();
-    
+
     assert!(
         result.is_ok(),
         "Unicode in regex pattern should work: {:?}",
@@ -88,7 +88,7 @@ stdout "Tests: ✓ ✗ ⚠ ℹ"
     fs::write(&test_file, test_content).unwrap();
 
     let result = testscript::run(testdata_dir.to_string_lossy()).execute();
-    
+
     assert!(
         result.is_ok(),
         "Multiple Unicode characters should work: {:?}",
@@ -111,12 +111,8 @@ stdout "Status: 🎉 Success!"
     fs::write(&test_file, test_content).unwrap();
 
     let result = testscript::run(testdata_dir.to_string_lossy()).execute();
-    
-    assert!(
-        result.is_ok(),
-        "Emoji in output should work: {:?}",
-        result
-    );
+
+    assert!(result.is_ok(), "Emoji in output should work: {:?}", result);
 }
 
 #[test]
@@ -134,7 +130,7 @@ stdout "Result: ✓ \(success\)"
     fs::write(&test_file, test_content).unwrap();
 
     let result = testscript::run(testdata_dir.to_string_lossy()).execute();
-    
+
     assert!(
         result.is_ok(),
         "Unicode with regex special chars should work: {:?}",
@@ -157,10 +153,36 @@ stdout "A.B"
     fs::write(&test_file, test_content).unwrap();
 
     let result = testscript::run(testdata_dir.to_string_lossy()).execute();
-    
+
     assert!(
         result.is_ok(),
         "Dot in regex should match Unicode character: {:?}",
         result
     );
 }
+
+#[test]
+fn test_grep_with_unicode() {
+    let temp_dir = TempDir::new().unwrap();
+    let testdata_dir = temp_dir.path().join("testdata");
+    fs::create_dir(&testdata_dir).unwrap();
+
+    // Test grep command with Unicode content
+    let test_content = r#"grep "Success.*✓" test.txt
+stdout "test.txt:1: Success ✓"
+
+-- test.txt --
+Success ✓
+"#;
+
+    let test_file = testdata_dir.join("grep_unicode_test.txt");
+    fs::write(&test_file, test_content).unwrap();
+
+    let result = testscript::run(testdata_dir.to_string_lossy()).execute();
+
+    assert!(
+        result.is_ok(),
+        "Grep with Unicode should work: {:?}",
+        result
+    );
+}