trying to get unicode idents to work

6 files changed, 43 insertions, 12 deletions

diff --git a/base_cgen.c b/base_cgen.c
index 125146b..b13012c 100644
--- a/base_cgen.c
+++ b/base_cgen.c
@@ -99,7 +99,9 @@ static bool cgen_ident(CGenerator *g, Identifier i, Location *where) {
 		}
 	}
 	cgen_indent(g);
-	fprint_ident(cgen_writing_to(g), i);
+	fprint_ident(stdout, i);
+	printf("\n");
+	fprint_ident_ascii(cgen_writing_to(g), i);
 	return true;
 }
 
diff --git a/identifiers.c b/identifiers.c
index b7a2b3c..93c7eb6 100644
--- a/identifiers.c
+++ b/identifiers.c
@@ -72,6 +72,17 @@ static void idents_create(Identifiers *ids) {
 	ids->root = ident_new(ids, NULL, 0); /* create root tree */
 }
 
+#if CHAR_MIN < 0
+#define ident_char_to_uchar(c) ((c) < 0 ? (256 + (c)) : (c))
+#else
+#define ident_char_to_uchar(c) (c)
+#endif
+
+#if CHAR_MIN < 0
+#define ident_uchar_to_char(c) ((c) > 127 ? ((c) - 256) : (c))
+#else
+#define ident_uchar_to_char(c) (c)
+#endif
 
 /* moves s to the char after the identifier */
 /* inserts if does not exist. reads until non-ident char is found. */
@@ -82,10 +93,11 @@ static Identifier ident_insert(Identifiers *ids, char **s) {
 		if (!isident(**s)) {
 			return tree;
 		}
-		int c = (**s) - CHAR_MIN;
+		int c = ident_char_to_uchar(**s);
 		assert(c >= 0 && c <= 255);
 		unsigned char c_low = (unsigned char)(c & 0xf);
 		unsigned char c_high = (unsigned char)(c >> 4);
+		printf("inserting %d as %d = %d, %d\n", **s, c, c_low, c_high);
 		if (!tree->children[c_low]) {
 			tree->children[c_low] = ident_new(ids, tree, c_low);
 		}
@@ -106,10 +118,27 @@ static void fprint_ident(FILE *out, Identifier id) {
 	fprint_ident(out, id->parent->parent); /* to go up one character, we need to go to the grandparent */
 	int c_low = id->parent->index_in_parent;
 	int c_high = id->index_in_parent;
-	int c = c_low + (c_high << 4) + CHAR_MIN; 
+	int c = ident_uchar_to_char(c_low + (c_high << 4)); 
 	fputc(c, out);
 }
 
+static void fprint_ident_ascii(FILE *out, Identifier id) {
+	assert(id);
+	if (id->parent == NULL) return; /* at root */
+	fprint_ident(out, id->parent->parent); /* to go up one character, we need to go to the grandparent */
+	int c_low = id->parent->index_in_parent;
+	int c_high = id->index_in_parent;
+	int c = c_low + (c_high << 4);
+	printf("Got %d as %d, %d\n", c, c_low, c_high);
+	if (c > 127) {
+		puts("x thing");
+		fprintf(out, "x__%x",c);
+	} else {
+		printf("single char %d\n",c);
+		fputc(ident_uchar_to_char(c), out);
+	}
+}
+
 /* NULL = no such identifier */
 static Identifier ident_get(Identifiers *ids, const char *s) {
 	IdentTree *tree = ids->root;
@@ -137,7 +166,7 @@ static char *ident_to_str(Identifier i) {
 		str--;
 		unsigned char c_high = i->index_in_parent;
 		unsigned char c_low = i->parent->index_in_parent;
-		char c = (char)(CHAR_MIN + (int)(c_low + (c_high << 4)));
+		char c = (char)ident_uchar_to_char((int)c_low + ((int)c_high << 4));
 		*str = c;
 		i = i->parent->parent; /* go to grandparent (prev char) */
 	}
diff --git a/out.c b/out.c
index d363960..49ba6b7 100644
--- a/out.c
+++ b/out.c
@@ -1,10 +1,10 @@
 #include "out.h"
 
 /* toc */
-void foo(void) {
+void Äx__a8(void) {
 }
-void main__(void) {
-	foo();
+void main(void) {
+	Äx__a8();
 }
 
 int main(void) {
diff --git a/out.h b/out.h
index 8d1ba2b..23bf38f 100644
--- a/out.h
+++ b/out.h
@@ -1,4 +1,4 @@
 #include <stddef.h>
 #include <stdint.h>
-void foo(void);
-void main__(void);
+void Äx__a8(void);
+void main(void);
diff --git a/test.toc b/test.toc
index 10e160d..4e46515 100644
--- a/test.toc
+++ b/test.toc
@@ -1,5 +1,6 @@
-foo @= fn() {
+Ĩ @= fn() {
 };
 main @= fn() {
-	 foo();
+	 Ĩ();
+
 };
diff --git a/util/err.c b/util/err.c
index 1a05e9d..72e94be 100644
--- a/util/err.c
+++ b/util/err.c
@@ -52,7 +52,6 @@ static void warn_print_header_(LineNo line) {
 }
 
 static void err_print_footer_(const char *context) {
-	err_fprint("\n\there --> ");
 	const char *end = strchr(context, '\n');
 	int has_newline = end != NULL;
 	if (!has_newline)