Show offending lookahead chars when pretty-printing trees w/ errors

include/tree_sitter/parser/lr_parser.h

@@ -191,6 +191,8 @@ ts_lr_parser_tree_root(ts_lr_parser *parser) {
     ts_tree *top_node = ts_stack_top_node(stack);
     if (stack->size <= 1)
         return top_node;
+    if (ts_tree_symbol(top_node) == ts_builtin_sym_error)
+        return top_node;
 
     size_t immediate_child_count;
     ts_tree **immedate_children = ts_tree_immediate_children(top_node, &immediate_child_count);

spec/runtime/languages/arithmetic/errors.txt

@@ -3,7 +3,7 @@ recovers from errors at the top level
 =====================================================
 x * * y
 ---
-(ERROR)
+(ERROR '*')
 
 =====================================================
 recovers from errors inside parenthesized expressions
@@ -12,4 +12,4 @@ x + (y * + z) * 5
 ---
 (expression (sum
     (variable)
-    (product (group (ERROR)) (number))))
+    (product (group (ERROR '+')) (number))))

spec/runtime/languages/javascript/errors.txt

@@ -3,7 +3,7 @@ recovers from errors in function calls
 ==========================================
 stuff(|||);
 ---
-(program (expression_statement (function_call (identifier) (ERROR))))
+(program (expression_statement (function_call (identifier) (ERROR '|'))))
 
 ==========================================
 recovers from errors in if statements
@@ -16,8 +16,8 @@ moreStuff();
 ---
 (program
     (expression_statement (function_call (identifier)))
-    (if_statement (ERROR)
-        (statement_block (expression_statement (ERROR))))
+    (if_statement (ERROR '*')
+        (statement_block (expression_statement (ERROR '*'))))
     (expression_statement (function_call (identifier))))
 
 ==========================================
@@ -33,8 +33,8 @@ moreStuff();
     (expression_statement (function_call (identifier)))
     (for_statement
         (var_declaration (assignment (identifier) (number)))
-        (expression_statement (ERROR))
-        (ERROR)
-        (statement_block (expression_statement (ERROR))))
+        (expression_statement (ERROR '*'))
+        (ERROR '*')
+        (statement_block (expression_statement (ERROR '*'))))
     (expression_statement (function_call (identifier))))
 

spec/runtime/languages/json/errors.txt

@@ -3,14 +3,14 @@ recovers from top-level errors
 ==========================================
 [}
 ---
-(ERROR)
+(ERROR '}')
 
 ==========================================
 recovers from unexpected tokens
 ==========================================
 barf
 ---
-(ERROR)
+(ERROR 'b')
 
 ==========================================
 recovers from errors inside arrays
@@ -19,7 +19,7 @@ recovers from errors inside arrays
 ---
 (value (array
     (number)
-    (ERROR)
+    (ERROR <EOF>)
     (number)))
 
 ==========================================
@@ -27,7 +27,7 @@ recovers from errors inside objects
 ==========================================
 { "key1": 1, oops }
 ---
-(value (object (string) (number) (ERROR)))
+(value (object (string) (number) (ERROR 'o')))
 
 ==========================================
 recovers from errors inside nested objects
@@ -35,6 +35,6 @@ recovers from errors inside nested objects
 { "key1": { "key2": 1, 2 }, [, "key3": 3 }
 ---
 (value (object
-    (string) (object (string) (number) (ERROR))
-    (ERROR)
+    (string) (object (string) (number) (ERROR '2'))
+    (ERROR '[')
     (string) (number)))

src/runtime/tree.c

@@ -124,14 +124,27 @@ ts_tree ** ts_tree_children(const ts_tree *tree, size_t *count) {
     return tree->data.children.contents;
 }
 
+static size_t write_lookahead_to_string(char *string, size_t limit, char lookahead) {
+    switch (lookahead) {
+        case '\0':
+            return snprintf(string, limit, "<EOF>");
+        default:
+            return snprintf(string, limit, "'%c'", lookahead);
+    }
+}
+
 static size_t tree_write_to_string(const ts_tree *tree, const char **symbol_names, char *string, size_t limit) {
     char *cursor = string;
     char **destination = (limit > 0) ? &cursor : &string;
 
     if (!tree)
         return snprintf(*destination, limit, "(NULL)");
-    if (tree->symbol == ts_builtin_sym_error)
-        return snprintf(*destination, limit, "(ERROR)");
+    if (tree->symbol == ts_builtin_sym_error) {
+        cursor += snprintf(*destination, limit, "(ERROR ");
+        cursor += write_lookahead_to_string(*destination, limit, tree->data.error.lookahead_char);
+        cursor += snprintf(*destination, limit, ")");
+        return cursor - string;
+    }
 
     cursor += snprintf(*destination, limit, "(%s", symbol_names[tree->symbol]);
     size_t count;