From d463acdbee5bb3a9509cb8414602f495408583b4 Mon Sep 17 00:00:00 2001
From: Tyge Løvset <tylo@norceresearch.no>
Date: Thu, 9 Jun 2022 16:59:52 +0200
Subject: Added utf8_peek()

---
 src/utf8code.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

(limited to 'src/utf8code.c')

diff --git a/src/utf8code.c b/src/utf8code.c
index 9613ba95..1c1e4336 100644
--- a/src/utf8code.c
+++ b/src/utf8code.c
@@ -45,6 +45,16 @@ unsigned utf8_encode(char *out, uint32_t c)
     return 0;
 }
 
+uint32_t utf8_peek(const char* s, int pos) {
+    int inc = 1;
+    if (pos < 0) pos = -pos, inc = -1;
+    while (pos)
+        pos -= (*(s += inc) & 0xC0) != 0x80;
+    utf8_decode_t d = {.state=0};
+    do { utf8_decode(&d, (uint8_t)*s++); } while (d.state);
+    return d.codep;
+}
+
 bool utf8_valid(const char* s) {
     utf8_decode_t d = {.state=0};
     while (*s)
-- 
cgit v1.2.3