Add explicit Unicode mode to regex patterns for proper Unicode handling (#40)

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)))?;
 

testdata/unicode_dot_regex.txt

@@ -0,0 +1,3 @@
+# Test if . in regex matches Unicode characters
+exec echo "A✓B"
+stdout "A.B"

testdata/unicode_emoji.txt

@@ -0,0 +1,3 @@
+# Test emoji in output
+exec echo "Status: 🎉 Success!"
+stdout "Status: 🎉 Success!"

testdata/unicode_grep.txt

@@ -0,0 +1,6 @@
+# Test grep command with Unicode content
+grep "Success.*✓" test.txt
+stdout "test.txt:1: Success ✓"
+
+-- test.txt --
+Success ✓

testdata/unicode_multiple.txt

@@ -0,0 +1,3 @@
+# Test multiple Unicode characters
+exec echo "Tests: ✓ ✗ ⚠ ℹ"
+stdout "Tests: ✓ ✗ ⚠ ℹ"

testdata/unicode_regex.txt

@@ -0,0 +1,3 @@
+# Test Unicode in regex pattern
+exec echo "Success ✓ completed"
+stdout "Success .* completed"

testdata/unicode_special_chars.txt

@@ -0,0 +1,3 @@
+# Test Unicode with regex special characters
+exec echo "Result: ✓ (success)"
+stdout "Result: ✓ \(success\)"

testdata/unicode_stderr.txt

@@ -0,0 +1,3 @@
+# Test Unicode checkmark in stderr exact match
+exec sh -c 'echo "Error ✗" >&2'
+stderr "Error ✗"

testdata/unicode_stdout.txt

@@ -0,0 +1,3 @@
+# Test Unicode checkmark in stdout exact match
+exec echo "Success ✓"
+stdout "Success ✓"