Temporarily fix small issue in parser

Change the way unary signs are handled (e.g., "+1e9", "-15"):
- instead of lexer, it is not the job of the parser
- due to how positional arguments chain together, expressions now need
  to be grouped using curly brackets in positional arguments list in
  order to avoid ambiguous netlists (e.g., `wg1 in out 100e-9 +2`)
- `-` appearing in net names could also pose issue but I think it's ok
  now ?

Now:
- `wg1 in out {2 - 1}` → one positional arg (= 2-1 = 1)
- `wg1 in out 2 - 1` → two positional args (2 and -1)
- `wg1 in out-abc + 1` → one positional args (1) (output net name: `out-abc`)
- `wg1 in out -  abc + 1` → one positional args (1) (output net name: `out-abc`)

Ultimately this should be revamped in a better way, but the way the parser works,
it was the best minimal change fix. It should hold for now.

1 files changed, 4 insertions, 2 deletions

diff --git a/src/parser/parser.l b/src/parser/parser.l
index 3519d0d..ad86569 100644
--- a/src/parser/parser.l
+++ b/src/parser/parser.l
@@ -336,13 +336,15 @@ string convert_special_char(char c)
     return '"';
     }
 
-[+-]?{D}+ {
+(?# [+-]?{D}+)
+{D}+ {
     /* signed integer number */
     yylval_param->i_val = from_string<int>(yytext);
     return T_INT;
 }
 
-[+-]?{DOUBLE} {
+(?# [+-]?{DOUBLE}+)
+{DOUBLE} {
     /* signed floating point number */
     yylval_param->d_val = from_string<double>(yytext);
     return T_NUM;