ww source: align UTF-8 BOM placement

cmd/wcc/lex.c

@@ -7,6 +7,15 @@
 #include <string.h>
 #include <errno.h>
 
+static int
+bomat(const char *src, u64 len, u64 pos)
+{
+	return pos <= len && len - pos >= 3
+	    && (unsigned char)src[pos] == 0xef
+	    && (unsigned char)src[pos + 1] == 0xbb
+	    && (unsigned char)src[pos + 2] == 0xbf;
+}
+
 void
 lexinit(Lex *l, Arena *a, const char *file, const char *src, u64 len)
 {
@@ -17,6 +26,12 @@ lexinit(Lex *l, Arena *a, const char *file, const char *src, u64 len)
 	l->line = 1;
 	l->col = 1;
 	l->a = a;
+	/* Go 1.26.5 syntax.source.nextch ignores U+FEFF only at the first
+	 * source position but counts its three bytes for the next column. */
+	if (bomat(src, len, 0)) {
+		l->pos = 3;
+		l->col = 4;
+	}
 }
 
 static int
@@ -25,6 +40,8 @@ lpeek(Lex *l, u64 ahead)
 	u64 p = l->pos + ahead;
 	if (p >= l->srclen)
 		return -1;
+	if (bomat(l->src, l->srclen, p))
+		return 0xfeff;
 	return (unsigned char)l->src[p];
 }
 
@@ -33,6 +50,14 @@ lget(Lex *l)
 {
 	if (l->pos >= l->srclen)
 		return -1;
+	if (bomat(l->src, l->srclen, l->pos)) {
+		Pos p = { l->file, l->line, l->col };
+		l->pos += 3;
+		l->col += 3;
+		errorf(p, "invalid BOM in the middle of the file");
+		l->errs++;
+		return 0xfeff;
+	}
 	int c = (unsigned char)l->src[l->pos++];
 	if (c == '\n') {
 		l->line++;
@@ -529,6 +554,11 @@ lexnext(Lex *l)
 		return t;
 	}
 	int c = lpeek(l, 0);
+	if (c == 0xfeff) {
+		lget(l);
+		Tok t = (Tok){ TK_ERR, start, "", 0, {0}, TK_NONE };
+		return t;
+	}
 
 	if (isidstart(c))
 		return lexident(l, start);

cmd/ww/main.c

@@ -4740,7 +4740,18 @@ sep_emit_body(FILE *out, const char *path, const char *modpath)
 	} else if (fputs("//ww:module-reset\n", out) == EOF) {
 		bad = 1;
 	}
-	if (fwrite(buf, 1, (size_t)len, out) != (size_t)len
+	/* Each physical source owns its own first-codepoint position. The
+	 * compiler sees one synthetic aggregate, so replace the optional UTF-8
+	 * BOM with position-preserving whitespace rather than turning a later
+	 * source's marker into a mid-unit BOM. */
+	u64 off = len >= 3
+	    && (unsigned char)buf[0] == 0xef
+	    && (unsigned char)buf[1] == 0xbb
+	    && (unsigned char)buf[2] == 0xbf ? 3 : 0;
+	if (off != 0 && fwrite("   ", 1, 3, out) != 3)
+		bad = 1;
+	if (fwrite(buf + off, 1, (size_t)(len - off), out)
+	    != (size_t)(len - off)
 	    || fputc('\n', out) == EOF)
 		bad = 1;
 	free(buf);