From a3da112d94058194e33d69d800293178605abc13 Mon Sep 17 00:00:00 2001
From: David Woodhouse <David.Woodhouse@intel.com>
Date: Wed, 4 Feb 2015 11:53:47 +0000
Subject: [PATCH] Split out get_utf8char() from buf_append_utf16le()

We'll want this for working around Yubikey/Android PIN weirdness.

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
---
 http.c                 | 89 ++++++++++++++++++++++--------------------
 openconnect-internal.h |  1 +
 2 files changed, 48 insertions(+), 42 deletions(-)

diff --git a/http.c b/http.c
index a55bebcb..8a459d9b 100644
--- a/http.c
+++ b/http.c
@@ -171,56 +171,61 @@ void buf_append_from_utf16le(struct oc_text_buf *buf, const void *_utf16)
 	buf_append_bytes(buf, utf8, 1);
 }
 
-int buf_append_utf16le(struct oc_text_buf *buf, const char *utf8)
+int get_utf8char(const char **p)
 {
-	int len = 0;
+	const char *utf8 = *p;
 	unsigned char c;
-	unsigned long utfchar;
-	int min;
-	int nr_extra;
+	int utfchar, nr_extra, min;
+
+	c = *(utf8++);
+	if (c < 128) {
+		utfchar = c;
+		nr_extra = 0;
+		min = 0;
+	} else if ((c & 0xe0) == 0xc0) {
+		utfchar = c & 0x1f;
+		nr_extra = 1;
+		min = 0x80;
+	} else if ((c & 0xf0) == 0xe0) {
+		utfchar = c & 0x0f;
+		nr_extra = 2;
+		min = 0x800;
+	} else if ((c & 0xf8) == 0xf0) {
+		utfchar = c & 0x07;
+		nr_extra = 3;
+		min = 0x10000;
+	} else {
+		return -EILSEQ;
+	}
 
-	/* Ick. Now I'm implementing my own UTF8 handling too. Perhaps it's
-	   time to bite the bullet and start requiring something like glib? */
-	while (*utf8) {
+	while (nr_extra--) {
 		c = *(utf8++);
-		if (c < 128) {
-			utfchar = c;
-			nr_extra = 0;
-			min = 0;
-		} else if ((c & 0xe0) == 0xc0) {
-			utfchar = c & 0x1f;
-			nr_extra = 1;
-			min = 0x80;
-		} else if ((c & 0xf0) == 0xe0) {
-			utfchar = c & 0x0f;
-			nr_extra = 2;
-			min = 0x800;
-		} else if ((c & 0xf8) == 0xf0) {
-			utfchar = c & 0x07;
-			nr_extra = 3;
-			min = 0x10000;
-		} else {
-			if (buf)
-				buf->error = -EILSEQ;
+		if ((c & 0xc0) != 0x80)
 			return -EILSEQ;
-		}
 
-		while (nr_extra--) {
-			c = *(utf8++);
-			if ((c & 0xc0) != 0x80) {
-				if (buf)
-					buf->error = -EILSEQ;
-				return -EILSEQ;
-			}
-			utfchar <<= 6;
-			utfchar |= (c & 0x3f);
-		}
-		if (utfchar > 0x10ffff || utfchar < min) {
+		utfchar <<= 6;
+		utfchar |= (c & 0x3f);
+	}
+	if (utfchar > 0x10ffff || utfchar < min)
+		return -EILSEQ;
+
+	*p = utf8;
+	return utfchar;
+}
+
+int buf_append_utf16le(struct oc_text_buf *buf, const char *utf8)
+{
+	int utfchar, len = 0;
+
+	/* Ick. Now I'm implementing my own UTF8 handling too. Perhaps it's
+	   time to bite the bullet and start requiring something like glib? */
+	while (*utf8) {
+		utfchar = get_utf8char(&utf8);
+		if (utfchar < 0) {
 			if (buf)
-				buf->error = -EILSEQ;
-			return -EILSEQ;
+				buf->error = utfchar;
+			return utfchar;
 		}
-
 		if (!buf)
 			continue;
 
diff --git a/openconnect-internal.h b/openconnect-internal.h
index a71211f6..b65c01da 100644
--- a/openconnect-internal.h
+++ b/openconnect-internal.h
@@ -920,6 +920,7 @@ void  __attribute__ ((format (printf, 2, 3)))
 void buf_append_bytes(struct oc_text_buf *buf, const void *bytes, int len);
 void buf_append_base64(struct oc_text_buf *buf, const void *bytes, int len);
 int buf_append_utf16le(struct oc_text_buf *buf, const char *utf8);
+int get_utf8char(const char **utf8);
 void buf_append_from_utf16le(struct oc_text_buf *buf, const void *utf16);
 void *openconnect_base64_decode(int *len, const char *in);
 void buf_truncate(struct oc_text_buf *buf);