fix(builtins): resolve 10 eval-surfaced interpreter bugs

crates/bashkit/src/builtins/awk.rs

@@ -1889,14 +1889,41 @@ impl AwkInterpreter {
                 }
                 let s = self.eval_expr(&args[0]).as_string();
                 let pattern = self.eval_expr(&args[1]).as_string();
+                // Extract capture array name from 3rd arg (gawk extension)
+                let arr_name = if args.len() >= 3 {
+                    if let AwkExpr::Variable(name) = &args[2] {
+                        Some(name.clone())
+                    } else {
+                        None
+                    }
+                } else {
+                    None
+                };
                 if let Ok(re) = Regex::new(&pattern) {
-                    if let Some(m) = re.find(&s) {
+                    if let Some(caps) = re.captures(&s) {
+                        let m = caps.get(0).unwrap();
                         let rstart = m.start() + 1; // awk is 1-indexed
                         let rlength = m.end() - m.start();
                         self.state
                             .set_variable("RSTART", AwkValue::Number(rstart as f64));
                         self.state
                             .set_variable("RLENGTH", AwkValue::Number(rlength as f64));
+                        // Populate capture array if 3rd arg provided
+                        if let Some(ref arr) = arr_name {
+                            // arr[0] = entire match
+                            let full_key = format!("{}[0]", arr);
+                            self.state
+                                .set_variable(&full_key, AwkValue::String(m.as_str().to_string()));
+                            // arr[1..N] = capture groups
+                            for i in 1..caps.len() {
+                                let key = format!("{}[{}]", arr, i);
+                                let val = caps
+                                    .get(i)
+                                    .map(|c| c.as_str().to_string())
+                                    .unwrap_or_default();
+                                self.state.set_variable(&key, AwkValue::String(val));
+                            }
+                        }
                         AwkValue::Number(rstart as f64)
                     } else {
                         self.state.set_variable("RSTART", AwkValue::Number(0.0));

crates/bashkit/src/builtins/cuttr.rs

@@ -202,31 +202,91 @@ impl Builtin for Tr {
     }
 }
 
-/// Expand a character set specification like "a-z" into a list of characters
+/// Expand a character set specification like "a-z" into a list of characters.
+/// Supports POSIX character classes: [:lower:], [:upper:], [:digit:], [:alpha:], [:alnum:], [:space:]
 fn expand_char_set(spec: &str) -> Vec<char> {
     let mut chars = Vec::new();
-    let mut iter = spec.chars().peekable();
-
-    while let Some(c) = iter.next() {
-        if iter.peek() == Some(&'-') {
-            iter.next(); // consume '-'
-            if let Some(&end) = iter.peek() {
-                iter.next(); // consume end char
-                             // Expand range
-                let start = c as u32;
-                let end = end as u32;
-                for code in start..=end {
-                    if let Some(ch) = char::from_u32(code) {
-                        chars.push(ch);
+    let mut i = 0;
+    let bytes = spec.as_bytes();
+
+    while i < bytes.len() {
+        // Check for POSIX character class [:class:]
+        if bytes[i] == b'[' && i + 1 < bytes.len() && bytes[i + 1] == b':' {
+            if let Some(end) = spec[i + 2..].find(":]") {
+                let class_name = &spec[i + 2..i + 2 + end];
+                match class_name {
+                    "lower" => chars.extend('a'..='z'),
+                    "upper" => chars.extend('A'..='Z'),
+                    "digit" => chars.extend('0'..='9'),
+                    "alpha" => {
+                        chars.extend('a'..='z');
+                        chars.extend('A'..='Z');
+                    }
+                    "alnum" => {
+                        chars.extend('a'..='z');
+                        chars.extend('A'..='Z');
+                        chars.extend('0'..='9');
+                    }
+                    "space" => chars.extend([' ', '\t', '\n', '\r', '\x0b', '\x0c']),
+                    "blank" => chars.extend([' ', '\t']),
+                    "print" | "graph" => {
+                        for code in 0x20u8..=0x7e {
+                            chars.push(code as char);
+                        }
+                    }
+                    _ => {
+                        // Unknown class, treat literally
+                        chars.push('[');
+                        i += 1;
+                        continue;
                     }
                 }
-            } else {
-                // Trailing dash, treat literally
-                chars.push(c);
-                chars.push('-');
+                i += 2 + end + 2; // skip past [: + class + :]
+                continue;
             }
+        }
+
+        let c = bytes[i] as char;
+        // Check for range like a-z
+        if i + 2 < bytes.len() && bytes[i + 1] == b'-' {
+            let end = bytes[i + 2] as char;
+            let start = c as u32;
+            let end = end as u32;
+            for code in start..=end {
+                if let Some(ch) = char::from_u32(code) {
+                    chars.push(ch);
+                }
+            }
+            i += 3;
+        } else if i + 1 == bytes.len() - 1 && bytes[i + 1] == b'-' {
+            // Trailing dash
+            chars.push(c);
+            chars.push('-');
+            i += 2;
         } else {
+            // Handle escape sequences
+            if c == '\\' && i + 1 < bytes.len() {
+                match bytes[i + 1] {
+                    b'n' => {
+                        chars.push('\n');
+                        i += 2;
+                        continue;
+                    }
+                    b't' => {
+                        chars.push('\t');
+                        i += 2;
+                        continue;
+                    }
+                    b'\\' => {
+                        chars.push('\\');
+                        i += 2;
+                        continue;
+                    }
+                    _ => {}
+                }
+            }
             chars.push(c);
+            i += 1;
         }
     }
 
@@ -338,6 +398,29 @@ mod tests {
         assert_eq!(expand_char_set("0-2"), vec!['0', '1', '2']);
     }
 
+    #[test]
+    fn test_expand_char_class_lower() {
+        let lower = expand_char_set("[:lower:]");
+        assert_eq!(lower.len(), 26);
+        assert_eq!(lower[0], 'a');
+        assert_eq!(lower[25], 'z');
+    }
+
+    #[test]
+    fn test_expand_char_class_upper() {
+        let upper = expand_char_set("[:upper:]");
+        assert_eq!(upper.len(), 26);
+        assert_eq!(upper[0], 'A');
+        assert_eq!(upper[25], 'Z');
+    }
+
+    #[tokio::test]
+    async fn test_tr_char_class_lower_to_upper() {
+        let result = run_tr(&["[:lower:]", "[:upper:]"], Some("hello world\n")).await;
+        assert_eq!(result.exit_code, 0);
+        assert_eq!(result.stdout, "HELLO WORLD\n");
+    }
+
     #[test]
     fn test_parse_field_spec() {
         assert_eq!(parse_field_spec("1"), vec![1]);

crates/bashkit/src/builtins/grep.rs

@@ -51,6 +51,7 @@ struct GrepOptions {
     count_only: bool,
     files_with_matches: bool,
     fixed_strings: bool,
+    extended_regex: bool,
     only_matching: bool,
     word_regex: bool,
     quiet: bool,
@@ -78,6 +79,7 @@ impl GrepOptions {
             count_only: false,
             files_with_matches: false,
             fixed_strings: false,
+            extended_regex: false,
             only_matching: false,
             word_regex: false,
             quiet: false,
@@ -114,8 +116,8 @@ impl GrepOptions {
                         'o' => opts.only_matching = true,
                         'w' => opts.word_regex = true,
                         'F' => opts.fixed_strings = true,
-                        'E' => {} // Extended regex is default
-                        'P' => {} // Perl regex - regex crate supports most Perl features
+                        'E' => opts.extended_regex = true,
+                        'P' => opts.extended_regex = true, // Perl regex implies ERE
                         'q' => opts.quiet = true,
                         'x' => opts.whole_line = true,
                         'H' => opts.show_filename = true,
@@ -298,6 +300,11 @@ impl GrepOptions {
                 }
                 let pat = if self.fixed_strings {
                     regex::escape(p)
+                } else if !self.extended_regex {
+                    // BRE mode: convert to ERE for the regex crate
+                    // In BRE: ( ) are literal, \( \) are groups
+                    // In ERE/regex crate: ( ) are groups, \( \) are literal
+                    bre_to_ere(p)
                 } else {
                     p.clone()
                 };
@@ -335,6 +342,43 @@ impl GrepOptions {
     }
 }
 
+/// Convert a BRE (Basic Regular Expression) pattern to ERE for the regex crate.
+/// In BRE: ( ) { } are literal; \( \) \{ \} \+ \? \| are metacharacters.
+/// In ERE/regex crate: ( ) { } + ? | are metacharacters.
+fn bre_to_ere(pattern: &str) -> String {
+    let mut result = String::with_capacity(pattern.len());
+    let chars: Vec<char> = pattern.chars().collect();
+    let mut i = 0;
+
+    while i < chars.len() {
+        if chars[i] == '\\' && i + 1 < chars.len() {
+            match chars[i + 1] {
+                // BRE escaped metacharacters → ERE unescaped
+                '(' | ')' | '{' | '}' | '+' | '?' | '|' => {
+                    result.push(chars[i + 1]);
+                    i += 2;
+                }
+                // Other escapes pass through
+                _ => {
+                    result.push('\\');
+                    result.push(chars[i + 1]);
+                    i += 2;
+                }
+            }
+        } else if chars[i] == '(' || chars[i] == ')' || chars[i] == '{' || chars[i] == '}' {
+            // BRE literal chars → escape them for ERE
+            result.push('\\');
+            result.push(chars[i]);
+            i += 1;
+        } else {
+            result.push(chars[i]);
+            i += 1;
+        }
+    }
+
+    result
+}
+
 #[async_trait]
 impl Builtin for Grep {
     async fn execute(&self, ctx: Context<'_>) -> Result<ExecResult> {