From 57eb0c15a10dfcd9816b84166106a1aa32323325 Mon Sep 17 00:00:00 2001
From: Nicholas Noll <nbnoll@eml.cc>
Date: Wed, 20 May 2020 12:15:24 -0700
Subject: fix: made rune constants unsigned for implicit casting

---
 include/libn.h   | 22 ++++++++++++----------
 sys/cmd/cc/cc.c  |  2 ++
 sys/cmd/cc/cc.h  |  4 ++--
 sys/cmd/cc/lex.c | 17 ++++++++---------
 sys/cmd/cc/pp.c  |  5 ++---
 5 files changed, 26 insertions(+), 24 deletions(-)

diff --git a/include/libn.h b/include/libn.h
index f469c5c..835d2c1 100644
--- a/include/libn.h
+++ b/include/libn.h
@@ -136,22 +136,24 @@ void    str·replace(string s, const byte* from, const byte* to);
 string* str·split(string s, const byte* tok);
 string  str·join(vlong len, byte** fields, const byte* sep);
 
- /*
+/*
  * UTF-8 functions.
  * Perhaps break into own unit
  * TODO: Add to(upper|lower|title)
  */
 typedef uint32 rune;
 
-enum
-{
-    UTFmax    = 4,
-    RuneSync  = 0x80,
-    RuneSelf  = 0x80,
-    RuneErr   = 0xFFFD,
-    RuneMax   = 0x10FFFF,
-    RuneMask  = 0x1FFFFF,
-};
+/* 
+ * We have to use the preprocessor to ensure
+ * we have unsigned constants. Unfortunate...
+ */
+
+#define UTFmax   4
+#define RuneSync 0x80u
+#define RuneSelf 0x80u
+#define RuneErr  0xFFFDu
+#define RuneMax  0x10FFFFu
+#define RuneMask 0x1FFFFFu
 
 /* utf8 helpers */
 int     utf8·fullrune(byte *s, int n);
diff --git a/sys/cmd/cc/cc.c b/sys/cmd/cc/cc.c
index 2a61e89..6338050 100644
--- a/sys/cmd/cc/cc.c
+++ b/sys/cmd/cc/cc.c
@@ -111,6 +111,8 @@ openio(byte *path)
             return nil;
     }
 
+    // TODO: See if we have already loaded the file
+    //
     printf("OPENING PATH %s\n", path);
 
     if ((C.io - C.iostk) >= arrlen(C.iostk)-1) 
diff --git a/sys/cmd/cc/cc.h b/sys/cmd/cc/cc.h
index 8a9fd80..6ed2caa 100644
--- a/sys/cmd/cc/cc.h
+++ b/sys/cmd/cc/cc.h
@@ -255,8 +255,8 @@ struct Lexer
 /* lex.c functions */
 Token lex(Lexer *);
 
-byte  getbyte(Lexer *);
-byte  getnsbyte(Lexer *l);
+int   getbyte(Lexer *);
+int   getnsbyte(Lexer *l);
 rune  getrune(Lexer *);
 byte  ungetbyte(Lexer *);
 rune  ungetrune(Lexer *, rune r);
diff --git a/sys/cmd/cc/lex.c b/sys/cmd/cc/lex.c
index 6c57b3c..8ce4e79 100644
--- a/sys/cmd/cc/lex.c
+++ b/sys/cmd/cc/lex.c
@@ -25,13 +25,13 @@ puttok(Token tok)
 // -----------------------------------------------------------------------
 // simple wrappers
 
-byte
+int
 getbyte(Lexer *lx)
 {
     return bufio·getbyte(&lx->io->buf);
 }
 
-byte
+int
 getnsbyte(Lexer *lx)
 {
     int b;
@@ -246,7 +246,6 @@ lex(Lexer *lx)
 {
     int    b, n, f;
     vlong  v, _;
-    uint   u;
     rune   r;
     string s;
     double d;
@@ -260,7 +259,7 @@ GetByte:
 Dispatch:
     tok.pos.beg = lx->pos;
 
-    if (b >= RuneSelf || b == '_')
+    if ((b != EOF && b >= RuneSelf) || b == '_')
         goto Talpha;
     if (isalpha(b)) {
         if (b != 'L')
@@ -604,7 +603,6 @@ Dispatch:
         goto Return;
 
     Talpha:
-        u = b; 
         s = lx->buf;
         e = lx->buf + arrlen(lx->buf);
         for (;;) {
@@ -612,18 +610,18 @@ Dispatch:
                 errorat(lx->pos, "identifier too long for buffer: %s", s);
                 goto Nospace;
             }
-            if (u != EOF && u >= RuneSelf) {
+            if (b != EOF && b >= RuneSelf) {
                 ungetbyte(lx);
                 r = getrune(lx);
                 if (!utf8·isletter(r) && !utf8·isdigit(r) && r != 0xb7) {
                     errorat(lx->pos, "invalid identifier character %d", r);
                 }
                 s += utf8·runetochar(s, &r);
-            } else if (!isalnum(u) && u != '_') 
+            } else if (!isalnum(b) && b != '_') 
                 break;
             else 
-                *s++ = u;
-            u = getbyte(lx);
+                *s++ = b;
+            b = getbyte(lx);
         }
         *s = '\0';
         ungetbyte(lx);
@@ -639,6 +637,7 @@ Dispatch:
         sym = lookup(&lx->sym, tok.val.s);
         if (sym) {
             io = makeio();
+            printf("EXPANDING MACRO %s\n", sym->name);
             io->buf.end   += expandmacro(lx, sym, io->b);
             *io->buf.end++ = EOF;
             pushio(lx, io);
diff --git a/sys/cmd/cc/pp.c b/sys/cmd/cc/pp.c
index cd8e28d..57cd5e2 100644
--- a/sys/cmd/cc/pp.c
+++ b/sys/cmd/cc/pp.c
@@ -40,7 +40,7 @@ identdots(Lexer *lx, int *dots)
     byte  *s;
 
     s = ident(lx);
-    if (s != nil)
+    if (*s != '\0')
         return s;
 
     c = getnsbyte(lx);
@@ -65,7 +65,7 @@ defmacro(Lexer *lx, string name, string macro)
 {
     Sym *mac;
 
-    printf("DEFINING MACRO %s ON LINE %d, file %s\n", name, lx->pos.line, os·basename(lx->pos.path));
+    // printf("DEFINING MACRO %s ON LINE %d, file %s\n", name, lx->pos.line, os·basename(lx->pos.path));
     mac = define(&lx->sym, name, Smacro);
     mac->macro = macro;
 
@@ -429,7 +429,6 @@ ppund(Lexer *lx)
     intern(&s);
     lx->b = lx->buf;
 
-    printf("FORGETTING %s\n", s);
     err = forget(&lx->sym, s);
     if (err) 
         warnat(lx->pos, "attempting to undefine unrecognized symbol '%s'", s);
-- 
cgit v1.2.1