fix: many bug fixes to number parsing

1 files changed, 13 insertions, 10 deletions

diff --git a/sys/cmd/cc/lex.c b/sys/cmd/cc/lex.c
index c84ea68..fc9dc0f 100644
--- a/sys/cmd/cc/lex.c
+++ b/sys/cmd/cc/lex.c
@@ -37,6 +37,11 @@ getnsbyte(Lexer *lx)
     int b;
     b = getbyte(lx);
     for (;;) {
+        if (b == EOF) {
+            popio(lx);
+            b = getbyte(lx);
+            continue;
+        }
         if (b >= RuneSelf || !isspace(b))
             return b;
         if (b == '\n')
@@ -477,24 +482,23 @@ Dispatch:
         if (b == '.' || tolower(b) == 'e')
             goto Tflt;
     Tint:
-        r = b;
         n = 10;
         s = lx->buf;
         if (*s == '0') {
-            b = *++s;
             switch (b) {
             case 'x': n = 16; break;
             case 'b': n = 2;  break;
             case 'o': n = 8;  break;
-            default: --s;
-            }
-            if (s >= e) {
-                errorat(lx->pos, "number overflows lexer buffer");
-                goto Nospace;
+            default: goto Rint;
             }
+            lx->b = s;
+            /* reparse number, now with base info */
+            while (b = getbyte(lx), (isdigit(b) || ('a' <= b && b <= 'f') || ('A' <= b && b <= 'F') || b == '_'))
+                *lx->b++ = b;
         }
-
+    Rint:
         v = 0;
+        r = b;
         for (; s != lx->b ; s++) {
             b = *s;
             if (b == '_') continue;
@@ -593,7 +597,7 @@ Dispatch:
                 errorat(lx->pos, "identifier too long for buffer: %s", s);
                 goto Nospace;
             }
-            if (u >= RuneSelf) {
+            if (u != EOF && u >= RuneSelf) {
                 ungetbyte(lx);
                 r = getrune(lx);
                 if (!utf8·isletter(r) && !utf8·isdigit(r) && r != 0xb7) {
@@ -626,7 +630,6 @@ Dispatch:
         }
         goto Return;
 
-
     default:
         tok.kind = Anil;
         errorat(lx->pos, "invalid token, crashing");