From 89c6c3940520acec39f1c2aabe7cbfc942a6ac89 Mon Sep 17 00:00:00 2001
From: Lephenixnoir <sebastien.michelland@protonmail.com>
Date: Mon, 1 Aug 2022 11:27:03 +0100
Subject: [PATCH] stdio: support for UTF-8 %lc in printf()

---
 include/fxlibc/printf.h         |  8 ++------
 src/stdio/printf/format_usual.c | 22 +++++++++++++++++++++-
 src/stdio/printf/print.c        |  8 ++++++--
 3 files changed, 29 insertions(+), 9 deletions(-)

diff --git a/include/fxlibc/printf.h b/include/fxlibc/printf.h
index 828ebff..bac624e 100644
--- a/include/fxlibc/printf.h
+++ b/include/fxlibc/printf.h
@@ -85,12 +85,8 @@ struct __printf_format {
 	int16_t precision;
 
 	/*
-	** Size specifier for integers (%o, %x, %i, %d, %u), may be one of:
-	** (0)  char       (8-bit)
-	** (1)  short      (16-bit)
-	** (2)  int        (32-bit)
-	** (3)  long       (32-bit)
-	** (4)  long long  (64-bit)
+	** Size specifier for integers (%o, %x, %i, %d, %u), is equal to the
+	** sizeof() of the targeted type. Also used for %lc.
 	*/
 	uint8_t size;
 
diff --git a/src/stdio/printf/format_usual.c b/src/stdio/printf/format_usual.c
index cfcc5f3..7c8eb64 100644
--- a/src/stdio/printf/format_usual.c
+++ b/src/stdio/printf/format_usual.c
@@ -19,7 +19,27 @@ void __printf_format_c(
 	__printf_compute_geometry(opt, &g);
 
 	__printf_outn(out, ' ', g.left_spaces);
-	__printf_out(out, c);
+
+	/* When using %lc, output as UTF-8 */
+	if(opt->size != 4 || c <= 0x7f) {
+		__printf_out(out, c);
+	}
+	else if(c <= 0x7ff) {
+		__printf_out(out, 0xc0 | (c >> 6));
+		__printf_out(out, 0x80 | (c & 0x3f));
+	}
+	else if(c <= 0xffff) {
+		__printf_out(out, 0xe0 | (c >> 12));
+		__printf_out(out, 0x80 | ((c >> 6) & 0x3f));
+		__printf_out(out, 0x80 | (c & 0x3f));
+	}
+	else {
+		__printf_out(out, 0xf0 | (c >> 18));
+		__printf_out(out, 0x80 | ((c >> 12) & 0x3f));
+		__printf_out(out, 0x80 | ((c >> 6) & 0x3f));
+		__printf_out(out, 0x80 | (c & 0x3f));
+	}
+
 	__printf_outn(out, ' ', g.right_spaces);
 }
 
diff --git a/src/stdio/printf/print.c b/src/stdio/printf/print.c
index 7b0efc3..e3269ff 100644
--- a/src/stdio/printf/print.c
+++ b/src/stdio/printf/print.c
@@ -125,8 +125,8 @@ void __printf_flush(struct __printf_output *out)
 /* Parse format strings. */
 static struct __printf_format parse_fmt(char const * restrict *options_ptr)
 {
-	/* No options enabled by default, set the size to int */
-	struct __printf_format opt = { .size = sizeof(int), .precision = -1 };
+	/* No options enabled by default; default size is set at the end */
+	struct __printf_format opt = { .size = 0, .precision = -1 };
 
 	/* This function acts as a deterministic finite automaton */
 	enum {
@@ -252,6 +252,10 @@ int __printf(
 		opt = parse_fmt(&format);
 		opt.spec = *format++;
 
+		/* Set default size to 1 for %c, 4 otherwise */
+		if(opt.size == 0)
+			opt.size = (opt.spec == 'c') ? 1 : 4;
+
 		int id;
 		if(isupper(opt.spec))
 			id = opt.spec - 'A';