Correct the caret position for unbalanced square brackets

5 files changed, 23 insertions, 2 deletions

diff --git a/src/fish_tests.cpp b/src/fish_tests.cpp
index 60de9fca..c507aca3 100644
--- a/src/fish_tests.cpp
+++ b/src/fish_tests.cpp
@@ -492,6 +492,16 @@ static void test_tok()
         do_test(token.error == TOK_UNTERMINATED_SUBSHELL);
         do_test(token.error_offset == 4);
     }
+    
+    {
+        tok_t token;
+        tokenizer_t t(L"abc defg[hij (klm)", 0);
+        do_test(t.next(&token));
+        do_test(t.next(&token));
+        do_test(token.type == TOK_ERROR);
+        do_test(token.error == TOK_UNTERMINATED_SLICE);
+        do_test(token.error_offset == 4);
+    }
 
     /* Test redirection_type_for_string */
     if (redirection_type_for_string(L"<") != TOK_REDIRECT_IN) err(L"redirection_type_for_string failed on line %ld", (long)__LINE__);
diff --git a/src/parse_constants.h b/src/parse_constants.h
index 552085f8..40ccc636 100644
--- a/src/parse_constants.h
+++ b/src/parse_constants.h
@@ -134,6 +134,7 @@ enum parse_error_code_t
     //tokenizer errors
     parse_error_tokenizer_unterminated_quote,
     parse_error_tokenizer_unterminated_subshell,
+    parse_error_tokenizer_unterminated_slice,
     parse_error_tokenizer_unterminated_escape,
     parse_error_tokenizer_other,
 
diff --git a/src/parse_tree.cpp b/src/parse_tree.cpp
index b31a6503..6238ad06 100644
--- a/src/parse_tree.cpp
+++ b/src/parse_tree.cpp
@@ -908,6 +908,10 @@ void parse_ll_t::report_tokenizer_error(const tok_t &tok)
         case TOK_UNTERMINATED_SUBSHELL:
             parse_error_code = parse_error_tokenizer_unterminated_subshell;
             break;
+            
+        case TOK_UNTERMINATED_SLICE:
+            parse_error_code = parse_error_tokenizer_unterminated_slice;
+            break;
 
         case TOK_UNTERMINATED_ESCAPE:
             parse_error_code = parse_error_tokenizer_unterminated_escape;
diff --git a/src/tokenizer.cpp b/src/tokenizer.cpp
index e361f88b..b50673ea 100644
--- a/src/tokenizer.cpp
+++ b/src/tokenizer.cpp
@@ -163,6 +163,9 @@ void tokenizer_t::read_string()
     // up to 96 open parens, before we give up on good error reporting
     const size_t paran_offsets_max = 96;
     size_t paran_offsets[paran_offsets_max];
+    
+    // where the open bracket is
+    size_t offset_of_bracket = 0;
 
     const wchar_t * const start = this->buff;
     bool is_first = true;
@@ -219,7 +222,10 @@ void tokenizer_t::read_string()
                         case L'[':
                         {
                             if (this->buff != start)
+                            {
                                 mode = mode_array_brackets;
+                                offset_of_bracket = this->buff - this->orig_buff;
+                            }
                             break;
                         }
 
@@ -361,8 +367,7 @@ void tokenizer_t::read_string()
             case mode_array_brackets:
             case mode_array_brackets_and_subshell:
             {
-                size_t offset_of_bracket = 0;
-                TOK_CALL_ERROR(this, TOK_UNTERMINATED_SUBSHELL, SQUARE_BRACKET_ERROR, this->orig_buff + offset_of_bracket); // TOK_UNTERMINATED_SUBSHELL is a lie but nobody actually looks at it
+                TOK_CALL_ERROR(this, TOK_UNTERMINATED_SLICE, SQUARE_BRACKET_ERROR, this->orig_buff + offset_of_bracket);
                 break;
             }
                 
diff --git a/src/tokenizer.h b/src/tokenizer.h
index dca8e5c2..d88feef6 100644
--- a/src/tokenizer.h
+++ b/src/tokenizer.h
@@ -39,6 +39,7 @@ enum tokenizer_error
     TOK_ERROR_NONE,
     TOK_UNTERMINATED_QUOTE,
     TOK_UNTERMINATED_SUBSHELL,
+    TOK_UNTERMINATED_SLICE,
     TOK_UNTERMINATED_ESCAPE,
     TOK_OTHER
 }