feat: support extglob patterns

grammar.js

@@ -55,6 +55,7 @@ module.exports = grammar({
     $._concat,
     $.variable_name, // Variable name followed by an operator like '=' or '+='
     $.regex,
+    $.extglob_pattern,
     '}',
     ']',
     '<<',
@@ -264,10 +265,14 @@ module.exports = grammar({
     ),
 
     case_item: $ => seq(
-      optional('('),
-      field('value', $._literal),
-      repeat(seq('|', field('value', $._literal))),
-      ')',
+      choice(
+        seq(
+          optional('('),
+          field('value', choice($._literal, $.extglob_pattern)),
+          repeat(seq('|', field('value', choice($._literal, $.extglob_pattern)))),
+          ')',
+        ),
+      ),
       optional($._statements),
       prec(1, choice(
         field('termination', ';;'),
@@ -277,8 +282,8 @@ module.exports = grammar({
 
     last_case_item: $ => seq(
       optional('('),
-      field('value', $._literal),
-      repeat(seq('|', field('value', $._literal))),
+      field('value', choice($._literal, $.extglob_pattern)),
+      repeat(seq('|', field('value', choice($._literal, $.extglob_pattern)))),
       ')',
       optional($._statements),
       optional(prec(1, ';;')),

src/scanner.c

@@ -47,6 +47,7 @@ enum TokenType {
     CONCAT,
     VARIABLE_NAME,
     REGEX,
+    EXTGLOB_PATTERN,
     CLOSING_BRACE,
     CLOSING_BRACKET,
     HEREDOC_ARROW,
@@ -77,12 +78,9 @@ static inline void advance(TSLexer *lexer) { lexer->advance(lexer, false); }
 static inline void skip(TSLexer *lexer) { lexer->advance(lexer, true); }
 
 static inline bool in_error_recovery(const bool *valid_symbols) {
-    if (valid_symbols[HEREDOC_START] && valid_symbols[HEREDOC_BODY_END] &&
-        valid_symbols[FILE_DESCRIPTOR] && valid_symbols[EMPTY_VALUE] &&
-        valid_symbols[CONCAT] && valid_symbols[REGEX]) {
-        return true;
-    }
-    return false;
+    return valid_symbols[HEREDOC_START] && valid_symbols[HEREDOC_BODY_END] &&
+           valid_symbols[FILE_DESCRIPTOR] && valid_symbols[EMPTY_VALUE] &&
+           valid_symbols[CONCAT] && valid_symbols[REGEX];
 }
 
 static unsigned serialize(Scanner *scanner, char *buffer) {
@@ -458,6 +456,79 @@ static bool scan(Scanner *scanner, TSLexer *lexer, const bool *valid_symbols) {
         }
     }
 
+    if (valid_symbols[EXTGLOB_PATTERN]) {
+        // first skip ws, then check for ? * + @ !
+        while (iswspace(lexer->lookahead)) {
+            skip(lexer);
+        }
+
+        if (lexer->lookahead == '?' || lexer->lookahead == '*' ||
+            lexer->lookahead == '+' || lexer->lookahead == '@' ||
+            lexer->lookahead == '!') {
+            lexer->mark_end(lexer);
+            advance(lexer);
+
+            if (lexer->lookahead != '(') {
+                return false;
+            }
+
+            typedef struct {
+                bool done;
+                uint32_t paren_depth;
+                uint32_t bracket_depth;
+                uint32_t brace_depth;
+            } State;
+
+            State state = {false, 0, 0, 0};
+            while (!state.done) {
+                switch (lexer->lookahead) {
+                    case '\0':
+                        return false;
+                    case '(':
+                        state.paren_depth++;
+                        break;
+                    case '[':
+                        state.bracket_depth++;
+                        break;
+                    case '{':
+                        state.brace_depth++;
+                        break;
+                    case ')':
+                        if (state.paren_depth == 0) {
+                            state.done = true;
+                        }
+                        state.paren_depth--;
+                        break;
+                    case ']':
+                        if (state.bracket_depth == 0) {
+                            state.done = true;
+                        }
+                        state.bracket_depth--;
+                        break;
+                    case '}':
+                        if (state.brace_depth == 0) {
+                            state.done = true;
+                        }
+                        state.brace_depth--;
+                        break;
+                }
+
+                if (!state.done) {
+                    bool was_space = iswspace(lexer->lookahead);
+                    advance(lexer);
+                    if (!was_space) {
+                        lexer->mark_end(lexer);
+                    }
+                }
+            }
+
+            lexer->result_symbol = EXTGLOB_PATTERN;
+            return true;
+        }
+
+        return false;
+    }
+
     return false;
 }
 

test/corpus/statements.txt

@@ -622,7 +622,11 @@ esac
 case x in x) echo meow ;; esac
 
 case foo in
-    bar\ baz) : ;;
+  bar\ baz) : ;;
+esac
+
+case "$arg" in
+  *([0-9])([0-9])) echo "$arg"
 esac
 
 --------------------------------------------------------------------------------
@@ -706,7 +710,19 @@ esac
       (word)
       (command
         (command_name
-          (word))))))
+          (word)))))
+  (case_statement
+    (string
+      (simple_expansion
+        (variable_name)))
+    (case_item
+      (extglob_pattern)
+      (command
+        (command_name
+          (word))
+        (string
+          (simple_expansion
+            (variable_name)))))))
 
 ================================================================================
 Test commands