vis: improve insertion of verbatim characters via CTRL-V in insert mode

Recognized formats are:

 CTRL-V nnn                     decimal value nnn
 CTRL-V onnn  or  CTRL-V Onnn   octal value nnn
 CTRL-V xnn   or  CTRL-V Xnn    hex value nn
 CTRL-V unnnn                   Unicode codepoint nnnn
 CTRL-V Unnnnnnnn               Unicode codepoint nnnnnnnn

Leading zeros can be omitted, any illegal character for the given
format will be ignored and terminates the numerical code.

1 files changed, 52 insertions, 0 deletions

diff --git a/libutf.c b/libutf.c
new file mode 100644
index 0000000..5f6ff45
--- /dev/null
+++ b/libutf.c
@@ -0,0 +1,52 @@
+/* libutf8 © 2012-2015 Connor Lane Smith <cls@lubutu.com> */
+#include "libutf.h"
+
+int
+runelen(Rune r)
+{
+	if(r <= 0x7F)
+		return 1;
+	else if(r <= 0x07FF)
+		return 2;
+	else if(r <= 0xD7FF)
+		return 3;
+	else if(r <= 0xDFFF)
+		return 0; /* surrogate character */
+	else if(r <= 0xFFFD)
+		return 3;
+	else if(r <= 0xFFFF)
+		return 0; /* illegal character */
+	else if(r <= Runemax)
+		return 4;
+	else
+		return 0; /* rune too large */
+}
+
+int
+runetochar(char *s, const Rune *p)
+{
+	Rune r = *p;
+
+	switch(runelen(r)) {
+	case 1: /* 0aaaaaaa */
+		s[0] = r;
+		return 1;
+	case 2: /* 00000aaa aabbbbbb */
+		s[0] = 0xC0 | ((r & 0x0007C0) >>  6); /* 110aaaaa */
+		s[1] = 0x80 |  (r & 0x00003F);        /* 10bbbbbb */
+		return 2;
+	case 3: /* aaaabbbb bbcccccc */
+		s[0] = 0xE0 | ((r & 0x00F000) >> 12); /* 1110aaaa */
+		s[1] = 0x80 | ((r & 0x000FC0) >>  6); /* 10bbbbbb */
+		s[2] = 0x80 |  (r & 0x00003F);        /* 10cccccc */
+		return 3;
+	case 4: /* 000aaabb bbbbcccc ccdddddd */
+		s[0] = 0xF0 | ((r & 0x1C0000) >> 18); /* 11110aaa */
+		s[1] = 0x80 | ((r & 0x03F000) >> 12); /* 10bbbbbb */
+		s[2] = 0x80 | ((r & 0x000FC0) >>  6); /* 10cccccc */
+		s[3] = 0x80 |  (r & 0x00003F);        /* 10dddddd */
+		return 4;
+	default:
+		return 0; /* error */
+	}
+}