From c2c59495b3450934ddb8c95d8e8612ac582bc104 Mon Sep 17 00:00:00 2001
From: Zack Buhman <zack@buhman.org>
Date: Wed, 23 Aug 2023 19:34:09 -0700
Subject: [PATCH] lexer: add support for binary number literals

---
 grammar.txt | 15 +++++++++++----
 lexer.cpp   | 26 +++++++++++++++++++++++++-
 2 files changed, 36 insertions(+), 5 deletions(-)

diff --git a/grammar.txt b/grammar.txt
index a09e048..2b4fabf 100644
--- a/grammar.txt
+++ b/grammar.txt
@@ -6,12 +6,19 @@ unary          → ( "~" | "+" | "-" ) unary
 shift          → andl ( ( "<<" | ">>" ) andl )*
 andl           → orl ( "&" orl )*
 orl            → andl ( ( "|" | "^" ) andl )*
-primary        → NUMBER
+primary        → number
                | "(" expression ")"
 
-uimm8  → expression
-uimm19 → expression
-uimm25 → expression
+number → "%" base2-number
+       | "$" base16-number
+       | "0b" base2-number
+       | "0x" base16-number
+       | base10-number
+
+imm → ("#")? expression
+uimm8  → imm
+uimm19 → imm
+uimm25 → imm
 
 alu → and | or | xor | add | sub | ad2 | sr | rr | sl | rl | rl8
 
diff --git a/lexer.cpp b/lexer.cpp
index cbc28f9..7f44134 100644
--- a/lexer.cpp
+++ b/lexer.cpp
@@ -53,6 +53,19 @@ constexpr static N parse_number(const std::string_view s)
   return n;
 }
 
+struct bin_t {
+  constexpr static bool pred(const char c)
+  {
+    return c >= '0' && c <= '1';
+  }
+
+  template <typename N>
+  constexpr static N parse(const std::string_view s)
+  {
+    return parse_number<N, 2>(s);
+  }
+};
+
 struct dec_t {
   constexpr static bool pred(const char c)
   {
@@ -159,7 +172,6 @@ std::optional<token_t> lexer_t::lex_token()
     case '-': return {{pos, minus, lexeme()}};
     case '*': return {{pos, star, lexeme()}};
     case '/': return {{pos, slash, lexeme()}};
-    case '%': return {{pos, percent, lexeme()}};
     case '~': return {{pos, tilde, lexeme()}};
     case '&': return {{pos, ampersand, lexeme()}};
     case '|': return {{pos, bar, lexeme()}};
@@ -188,6 +200,13 @@ std::optional<token_t> lexer_t::lex_token()
 	return {{tmp, eol, lexeme()}};
       }
       break;
+    case '%':
+      if (bin_t::pred(peek())) {
+	start_ix += 1;
+	return {_number<bin_t>()};
+      } else {
+        return {{pos, percent, lexeme()}};
+      }
     case '$':
       if (hex_t::pred(peek())) {
 	start_ix += 1;
@@ -200,6 +219,11 @@ std::optional<token_t> lexer_t::lex_token()
 	  start_ix += 2;
 	  return {_number<hex_t>()};
 	}
+      } else if (match('b')) {
+	if (bin_t::pred(peek())) {
+	  start_ix += 2;
+	  return {_number<bin_t>()};
+	}
       }
       [[fallthrough]];
     default: