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.

3 files changed, 144 insertions, 7 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 */
+	}
+}
diff --git a/libutf.h b/libutf.h
new file mode 100644
index 0000000..4d3f91c
--- /dev/null
+++ b/libutf.h
@@ -0,0 +1,34 @@
+#ifndef LIBUTF_H
+#define LIBUTF_H
+
+/* libutf8 © 2012-2015 Connor Lane Smith <cls@lubutu.com> */
+#include <stddef.h>
+#include <stdint.h>
+
+#if __STDC_VERSION__ >= 201112L
+#include <uchar.h>
+#ifdef __STDC_UTF_32__
+#define RUNE_C INT32_C
+typedef char32_t Rune;
+#endif
+#endif
+
+#ifndef RUNE_C
+#ifdef INT32_C
+#define RUNE_C INT32_C
+typedef uint_least32_t Rune;
+#else
+#define RUNE_C(x) x##L
+typedef unsigned long Rune;
+#endif
+#endif
+
+#define UTFmax 6 /* maximum bytes per rune */
+
+#define Runeself 0x80             /* rune and utf are equal (<) */
+#define Runemax  RUNE_C(0x10FFFF) /* maximum rune value */
+
+int runelen(Rune r);
+int runetochar(char *s, const Rune *p);
+
+#endif
diff --git a/vis.c b/vis.c
index ffe45ce..339bfb0 100644
--- a/vis.c
+++ b/vis.c
@@ -39,6 +39,7 @@
 #include "text-objects.h"
 #include "util.h"
 #include "map.h"
+#include "libutf.h"
 
 /** global variables */
 static Editor *vis;         /* global editor instance, keeps track of all windows etc. */
@@ -1037,15 +1038,65 @@ static void prompt_backspace(const Arg *arg) {
 }
 
 static void insert_verbatim(const Arg *arg) {
-	int value = 0;
-	for (int i = 0; i < 3; i++) {
-		Key k = getkey();
-		if (k.str[0] < '0' || k.str[0] > '9')
+	int len = 0, count = 0, base;
+	Rune rune = 0;
+	Key key = getkey();
+	char buf[4], type = key.str[0];
+	switch (type) {
+	case 'o':
+	case 'O':
+		count = 3;
+		base = 8;
+		break;
+	case 'U':
+		count = 4;
+		/* fall through */
+	case 'u':
+		count += 4;
+		base = 16;
+		break;
+	case 'x':
+	case 'X':
+		count = 2;
+		base = 16;
+		break;
+	default:
+		if (type < '0' || type > '9')
 			return;
-		value = value * 10 + k.str[0] - '0';
+		rune = type - '0';
+		count = 2;
+		base = 10;
+		break;
+	}
+
+	while (count-- > 0) {
+		key = getkey();
+		int v = 0;
+		if (base == 8 && '0' <= key.str[0] && key.str[0] <= '7')
+			v = key.str[0] - '0';
+		else if ((base == 10 || base == 16) && '0' <= key.str[0] && key.str[0] <= '9')
+			v = key.str[0] - '0';
+		else if (base == 16 && 'a' <= key.str[0] && key.str[0] <= 'f')
+			v = 10 + key.str[0] - 'a';
+		else if (base == 16 && 'A' <= key.str[0] && key.str[0] <= 'F')
+			v = 10 + key.str[0] - 'A';
+		else
+			break;
+		rune = rune * base + v;
+	}
+
+	if (type == 'u' || type == 'U') {
+		len = runetochar(buf, &rune);
+	} else {
+		buf[0] = rune;
+		len = 1;
+	}
+
+	if (len > 0) {
+		size_t pos = view_cursor_get(vis->win->view);
+		editor_insert(vis, pos, buf, len);
+		view_cursor_to(vis->win->view, pos + len);
 	}
-	char v = value;
-	editor_insert(vis, view_cursor_get(vis->win->view), &v, 1);
 }
 
 static void quit(const Arg *arg) {