From 4d2b3e59a169a721e24f3ac576cf8bc81ff54699 Mon Sep 17 00:00:00 2001
From: Steve Tautonico <stautonico@gmail.com>
Date: Sat, 7 Oct 2023 03:53:34 -0400
Subject: [PATCH] Added octal literal support

---
 src/lexer.cpp   | 20 ++++++++++++++------
 tests/01_int.py |  4 ++++
 2 files changed, 18 insertions(+), 6 deletions(-)

diff --git a/src/lexer.cpp b/src/lexer.cpp
index f033d504..1487033d 100644
--- a/src/lexer.cpp
+++ b/src/lexer.cpp
@@ -260,7 +260,7 @@ static bool is_unicode_Lo_char(uint32_t c) {
     }
 
     void Lexer::eat_number() {
-        PK_LOCAL_STATIC const std::regex pattern("^(0x)?[0-9a-fA-F]+(\\.[0-9]+)?(L)?");
+        PK_LOCAL_STATIC const std::regex pattern("^(0[xo])?[0-9a-fA-F]+(\\.[0-9]+)?(L)?");
         std::smatch m;
 
         const char* i = token_start;
@@ -278,20 +278,28 @@ static bool is_unicode_Lo_char(uint32_t c) {
         }
 
         if(m[1].matched && m[2].matched){
-            SyntaxError("hex literal should not contain a dot");
+            SyntaxError("hex/octal literal should not contain a dot");
         }
 
         try{
             int base = 10;
             size_t size;
-            if (m[1].matched) base = 16;
+            if (m[1].matched) {
+                if (m[1].str() == "0o") base=8;
+                else base = 16;
+            }
             if (m[2].matched) {
                 PK_ASSERT(base == 10);
                 add_token(TK("@num"), Number::stof(m[0], &size));
             } else {
-                add_token(TK("@num"), (i64)std::stoll(m[0], &size, base));
+                // If we're base 8, chop off the "o"
+                std::string match = m[0].str();
+                if (base == 8) match.erase(1, 1);
+                add_token(TK("@num"), (i64)std::stoll(match, &size, base));
             }
-            PK_ASSERT((int)size == (int)m.length());
+            // HACK: We need to check length-1 for octal since python octals are "0o..." and c/c++ octals are "0..."
+            if (base == 8) {PK_ASSERT((int)size == (int)m.length()-1);}
+            else {PK_ASSERT((int)size == (int)m.length());}
         }catch(...){
             SyntaxError("invalid number literal");
         }
@@ -466,4 +474,4 @@ static bool is_unicode_Lo_char(uint32_t c) {
         return std::move(nexts);
     }
 
-}   // namespace pkpy
\ No newline at end of file
+}   // namespace pkpy
diff --git a/tests/01_int.py b/tests/01_int.py
index 07b8d793..f95264a4 100644
--- a/tests/01_int.py
+++ b/tests/01_int.py
@@ -5,6 +5,10 @@ assert 0x7fffffff == 2147483647
 # test 64-bit
 assert 2**60-1 + 546 - 0xfffffffffffff == 1148417904979477026
 
+# test oct literals
+assert 0o1234 == 668
+assert 0o17777777777 == 2147483647
+
 # test == != >= <= < >
 assert -1 == -1
 assert -1 != 1