From 7326866a6348d62f1a49e604e0526c39ccdc842f Mon Sep 17 00:00:00 2001
From: Kaz Kylheku <kaz@kylheku.com>
Date: Wed, 1 Feb 2017 22:01:43 -0800
Subject: parser: diagnose run-on symbols.

* parser.l (grammar): Add rules which capture two symbols
glued together, and diagnose as bad token. Of course a
legitimate symbol token can be divided into two that are glued
together. This rule is placed after the legitimate symbol
matching rule, so that if a token can be interpreted as a
single symbol token or as two, the first interpretation is
taken.
---
 parser.l | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/parser.l b/parser.l
index 39751ed6..569c4a79 100644
--- a/parser.l
+++ b/parser.l
@@ -409,6 +409,20 @@ UONLY   {U2}{U}|{U3}{U}{U}|{U4}{U}{U}{U}
   return SYMTOK;
 }
 
+<BRACED>{BTOK}{BTOK} |
+<NESTED>{NTOK}{NTOK} {
+  if (yy_top_state(yyscanner) == INITIAL
+      || yy_top_state(yyscanner) == QSILIT
+      || yy_top_state(yyscanner) == QWLIT)
+    yy_pop_state(yyscanner);
+
+  yyerrorf(yyg, lit("bad token: ~a"),
+                    string_own(utf8_dup_from(yytext)),
+                    nao);
+  yylval->lexeme = utf8_dup_from(yytext);
+  return SYMTOK;
+}
+
 <SPECIAL>\({WS}all{WS}\) {
   yy_pop_state(yyscanner);
   yylval->lineno = yyextra->lineno;
-- 
cgit v1.2.3