From: Keith Packard <keithp@keithp.com>
Date: Mon, 8 Jan 2018 21:46:17 +0000 (-0800)
Subject: altos/scheme: Allow unicode in lexer
X-Git-Tag: 1.8.5~1^2~2^2~4
X-Git-Url: https://git.gag.com/?a=commitdiff_plain;h=4b52fc6eea9a478cb3dd42dcd32c92838df39734;hp=f8a967959b2f5ca3486ab3422f30fe4ad4ba17a8;p=fw%2Faltos

altos/scheme: Allow unicode in lexer

This just passes any bytes with the high bit set through the system so
programs can include UTF-8 in strings and symbols. What the heck.

Signed-off-by: Keith Packard <keithp@keithp.com>
---

diff --git a/src/scheme/ao_scheme_read.c b/src/scheme/ao_scheme_read.c
index f9630d39..3575ff3f 100644
--- a/src/scheme/ao_scheme_read.c
+++ b/src/scheme/ao_scheme_read.c
@@ -186,8 +186,9 @@ lexc(FILE *in)
 			c = 0;
 			lex_class = ENDOFFILE;
 		} else {
-			c &= 0x7f;
-			lex_class = lex_classes[c];
+			lex_class = PRINTABLE;
+			if (c <= 0x7f)
+				lex_class = lex_classes[c];
 		}
 	} while (lex_class & IGNORE);
 	return c;
diff --git a/src/scheme/ao_scheme_string.c b/src/scheme/ao_scheme_string.c
index c49e1e32..2c6d0960 100644
--- a/src/scheme/ao_scheme_string.c
+++ b/src/scheme/ao_scheme_string.c
@@ -47,6 +47,8 @@ ao_scheme_string_alloc(int len)
 {
 	struct ao_scheme_string	*s;
 
+	if (len < 0)
+		return NULL;
 	s = ao_scheme_alloc(len + 2);
 	if (!s)
 		return NULL;
@@ -182,8 +184,8 @@ ao_scheme_string_write(FILE *out, ao_poly p, bool write)
 				fputs("\\\\", out);
 				break;
 			default:
-				if (c < ' ')
-					fprintf(out, "\\%03o", c);
+				if ((uint8_t) c < ' ')
+					fprintf(out, "\\%03o", (uint8_t) c);
 				else
 					putc(c, out);
 				break;