From 1454f04265b999d665e4820775a55107a8a29c09 Mon Sep 17 00:00:00 2001
From: Allen Webster <yoyo4thdimention@gmail.com>
Date: Wed, 30 Jun 2021 15:46:39 -0400
Subject: [PATCH] detect broken strings in the lexer

---
 source/md.h      |  8 ++++----
 source/md_impl.c | 15 +++------------
 2 files changed, 7 insertions(+), 16 deletions(-)

diff --git a/source/md.h b/source/md.h
index 5163586..e59f847 100644
--- a/source/md.h
+++ b/source/md.h
@@ -427,9 +427,9 @@ enum
     MD_TokenKind_Comment        = (1<<5),
     MD_TokenKind_Whitespace     = (1<<6),
     MD_TokenKind_Newline        = (1<<7),
-    MD_TokenKind_BrokenComment  = (1<<8),
-    MD_TokenKind_BrokenString   = (1<<9),
-    MD_TokenKind_BadCharacter   = (1<<10),
+    MD_TokenKind_BrokenComment       = (1<<8),
+    MD_TokenKind_BrokenStringLiteral = (1<<9),
+    MD_TokenKind_BadCharacter        = (1<<10),
 };
 
 enum
@@ -445,7 +445,7 @@ enum
                              MD_TokenKind_StringLiteral|
                              MD_TokenKind_Symbol),
     MD_TokenGroup_Error   = (MD_TokenKind_BrokenComment|
-                             MD_TokenKind_BrokenString|
+                             MD_TokenKind_BrokenStringLiteral|
                              MD_TokenKind_BadCharacter),
 };
 
diff --git a/source/md_impl.c b/source/md_impl.c
index 3844abe..722e69e 100644
--- a/source/md_impl.c
+++ b/source/md_impl.c
@@ -1183,10 +1183,7 @@ MD_TokenFromString(MD_String8 string)
             case '\'':
             case '`':
             {
-                token.kind = MD_TokenKind_StringLiteral;
-                
-                // TODO(allen): proposal:
-                // go see the proposal in the block comment lexer, same idea here?
+                token.kind = MD_TokenKind_BrokenStringLiteral;
                 
                 // determine delimiter setup
                 MD_u8 d = *at;
@@ -1212,6 +1209,7 @@ MD_TokenFromString(MD_String8 string)
                             // close condition
                             if (consecutive_d == 3){
                                 chop_n = 3;
+                                token.kind = MD_TokenKind_StringLiteral;
                                 break;
                             }
                         }
@@ -1246,6 +1244,7 @@ MD_TokenFromString(MD_String8 string)
                         if (*at == d){
                             at += 1;
                             chop_n = 1;
+                            token.kind = MD_TokenKind_StringLiteral;
                             break;
                         }
                         
@@ -1775,16 +1774,8 @@ MD_ParseOneNode(MD_String8 string, MD_u64 offset)
                                       label_name.outer_string.str - string.str);
             parsed_node->flags |= label_name.node_flags;
             
-            //- rjf: check for string literal errors
             if(label_name.kind == MD_TokenKind_StringLiteral)
             {
-                if(!_MD_TokenBoundariesAreBalanced(label_name))
-                {
-                    MD_String8 capped = MD_StringPrefix(label_name.outer_string, MD_UNTERMINATED_TOKEN_LEN_CAP);
-                    MD_Error *error = MD_MakeNodeError(parsed_node, MD_MessageKind_CatastrophicError,
-                                                       MD_PushStringF("Unterminated text literal \"%.*s\"", MD_StringExpand(capped)));
-                    MD_PushErrorToList(&result.errors, error);
-                }
             }
             
             //- rjf: check for unexpected reserved symbols