sob

commit 953439a2efe7bf8be200be0d384377d1d96f4d48
parent 4f2bc0ed04d5afc5dfd11651fe84747eb3ec55bb
Author: Hiltjo Posthuma <hiltjo@codemadness.org>
Date:   Fri, 10 Oct 2014 18:47:55 +0000

add support for characters that are bigger than 1 column

Diffstat:
M
README
 | 
4
+++-
M
sob.c
 | 
78
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++--------------

2 files changed, 67 insertions(+), 15 deletions(-)
diff --git a/README b/README
@@ -10,6 +10,8 @@ Dependencies
 Features
 --------
 - Small (in size and memory), not much dependencies.
+- UTF-8 input and output support.
+	- Support for characters that are bigger than 1 column.
 - Custom prompt (including color support).
 - Easy to write custom completion scripts or special actions.
 - Custom action on SIGWINCH (window resize).
@@ -23,7 +25,7 @@ Features
 
 Known issues
 ------------
-line yank doesn't work with xclip, but does work with xsel.
+- Line yank doesn't work with xclip, but does work with xsel.
 
 
 Author
diff --git a/sob.c b/sob.c
@@ -12,6 +12,9 @@
 #include <unistd.h>
 #include <termios.h>
 
+#define __XOPEN_SOURCE
+#include <wchar.h>
+
 #include "arg.h"
 char *argv0;
 
@@ -27,6 +30,8 @@ struct line {
 	size_t utflen;     /* length in characters */
 	size_t bytepos;    /* index position (in bytes) */
 	size_t utfpos;     /* pos in characters */
+	size_t colpos;     /* cursor position (in columns) */
+	size_t collen;     /* total length (in columns) */
 };
 
 static void   line_clear(void);
@@ -69,10 +74,11 @@ static void   setup(void);
 static void   sighandler(int);
 static void   usage(void);
 
+static size_t colw(const char *, size_t);
 static int    nonspace(int c);
 static size_t utf8len(const char *);
-static size_t utfprevn(const char *, size_t , size_t);
-static size_t utfnextn(const char *, size_t , size_t);
+static size_t utfprevn(const char *, size_t, size_t);
+static size_t utfnextn(const char *, size_t, size_t);
 static void   utfuntilchar(size_t *, size_t *, int (*)(int), int);
 
 static struct termios ttystate, ttysave;
@@ -92,6 +98,25 @@ nonspace(int c)
 }
 
 static size_t
+colw(const char *s, size_t max)
+{
+	size_t len = 0, i;
+	wchar_t w = 0;
+	int r;
+
+	for(i = 0; *s; s++, i++) {
+		if((*s & 0xc0) != 0x80) {
+			if((r = mbtowc(&w, s, i + 4 > max ? max - i : 4)) == -1)
+				break;
+			if((r = wcwidth(w)) == -1)
+				break;
+			len += r;
+		}
+	}
+	return len;
+}
+
+static size_t
 utf8len(const char *s)
 {
 	size_t i;
@@ -133,7 +158,7 @@ utfnextn(const char *s, size_t p, size_t n)
 	return 0;
 }
 
-/* b is byte start pos, u is utf pos, f is filter function,
+/* b is byte start pos, u is utf pos, c is column pos, f is filter function,
  * dir is -1 or +1 for prev or next */
 static void
 utfuntilchar(size_t *b, size_t *u, int (*f)(int), int dir)
@@ -165,12 +190,13 @@ utfuntilchar(size_t *b, size_t *u, int (*f)(int), int dir)
 static void
 line_inserttext(const char *s)
 {
-	size_t siz, len;
+	size_t siz, ulen, clen;
 
 	siz = strlen(s);
 	if(line.bytepos + siz + 1 > sizeof(line.line))
 		return;
-	len = utf8len(s);
+	clen = colw(s, siz);
+	ulen = utf8len(s);
 	/* append */
 	if(line.bytepos == line.bytesiz) {
 		memmove(&line.line[line.bytepos], s, siz);
@@ -184,7 +210,9 @@ line_inserttext(const char *s)
 	line.bytesiz += siz;
 	line.line[line.bytesiz + 1] = '\0';
 	line.utflen = utf8len(line.line);
-	line.utfpos += len;
+	line.utfpos += ulen;
+	line.colpos += clen;
+	line.collen = colw(line.line, line.bytesiz);
 	line_draw();
 }
 
@@ -202,6 +230,8 @@ line_set(const char *s)
 	line.bytepos = line.bytesiz;
 	line.utflen = utf8len(line.line);
 	line.utfpos = line.utflen;
+	line.collen = colw(line.line, line.bytesiz);
+	line.colpos = line.collen;
 }
 
 /* like mksh, toggle counting of escape codes in prompt with "\x01" */
@@ -237,7 +267,7 @@ line_draw(void)
 	fprintf(outfp, "\x1b[2J\x1b[H"); /* clear */
 	line_prompt();
 	fwrite(line.line, 1, line.bytesiz, outfp);
-	line_cursor_move(line.utfpos);
+	line_cursor_move(line.colpos);
 }
 
 static void
@@ -268,14 +298,16 @@ static void
 line_cursor_wordprev(void)
 {
 	line_getwordposprev(line.bytepos, line.utfpos, &line.bytepos, &line.utfpos);
-	line_cursor_move(line.utfpos);
+	line.colpos = colw(line.line, line.bytepos);
+	line_cursor_move(line.colpos);
 }
 
 static void
 line_cursor_wordnext(void)
 {
 	line_getwordposnext(line.bytepos, line.utfpos, &line.bytepos, &line.utfpos);
-	line_cursor_move(line.utfpos);
+	line.colpos = colw(line.line, line.bytepos);
+	line_cursor_move(line.colpos);
 }
 
 static void
@@ -283,7 +315,8 @@ line_cursor_begin(void)
 {
 	line.bytepos = 0;
 	line.utfpos = 0;
-	line_cursor_move(line.utfpos);
+	line.colpos = 0;
+	line_cursor_move(line.colpos);
 }
 
 static void
@@ -298,7 +331,8 @@ line_cursor_prev(void)
 
 	line.bytepos -= n;
 	line.utfpos--;
-	line_cursor_move(line.utfpos);
+	line.colpos -= colw(&line.line[line.bytepos], n);
+	line_cursor_move(line.colpos);
 }
 
 static void
@@ -311,9 +345,10 @@ line_cursor_next(void)
 
 	if((n = utfnextn(line.line, line.bytepos, 1)) == 0)
 		return;
+	line.colpos += colw(&line.line[line.bytepos], n);
 	line.bytepos += n;
 	line.utfpos++;
-	line_cursor_move(line.utfpos);
+	line_cursor_move(line.colpos);
 }
 
 static void
@@ -321,7 +356,8 @@ line_cursor_end(void)
 {
 	line.bytepos = line.bytesiz;
 	line.utfpos = line.utflen;
-	line_cursor_move(line.utfpos);
+	line.colpos = line.collen;
+	line_cursor_move(line.colpos);
 }
 
 static void
@@ -342,6 +378,9 @@ line_delcharnext(void)
 
 	if((siz = utfnextn(line.line, line.bytepos, 1)) == 0)
 		return;
+
+	line.collen -= colw(&line.line[line.bytepos], siz);
+
 	memmove(&line.line[line.bytepos], &line.line[line.bytepos + siz],
 	        line.bytesiz - line.bytepos - siz);
 
@@ -354,20 +393,25 @@ line_delcharnext(void)
 static void
 line_delcharprev(void)
 {
-	size_t siz;
+	size_t siz, col;
 
 	if(line.utfpos <= 0 || line.utflen <= 0)
 		return;
 	if((siz = utfprevn(line.line, line.bytepos, 1)) == 0)
 		return;
 
+	col = colw(&line.line[line.bytepos - siz], siz);
+
 	memmove(&line.line[line.bytepos - siz], &line.line[line.bytepos],
 	        line.bytesiz - line.bytepos);
+
 	line.bytepos -= siz;
 	line.bytesiz -= siz;
 	line.line[line.bytesiz] = '\0';
 	line.utflen--;
 	line.utfpos--;
+	line.colpos -= col;
+	line.collen -= col;
 
 	line_draw();
 }
@@ -379,6 +423,8 @@ line_deltoend(void)
 	line.bytesiz = line.bytepos;
 	line.utflen = utf8len(line.line);
 	line.utfpos = line.utflen;
+	line.collen = colw(line.line, line.bytesiz);
+	line.colpos = line.collen;
 
 	line_draw();
 }
@@ -400,6 +446,8 @@ line_delwordcursor(void)
 	line.line[line.bytesiz] = '\0';
 	line.utfpos -= len;
 	line.utflen -= len;
+	line.collen = colw(line.line, line.bytesiz);
+	line.colpos = colw(line.line, bs);
 
 	line_draw();
 }
@@ -415,6 +463,7 @@ line_delwordprev(void)
 	line_getwordposprev(line.bytepos, line.utfpos, &bs, &us);
 
 	siz = line.bytepos - bs;
+	line.colpos -= colw(&line.line[bs], siz);
 	memmove(&line.line[bs], &line.line[line.bytepos],
 	        line.bytesiz - line.bytepos);
 
@@ -425,6 +474,7 @@ line_delwordprev(void)
 	line.line[line.bytesiz] = '\0';
 	line.utfpos -= len;
 	line.utflen -= len;
+	line.collen = colw(line.line, line.bytesiz);
 
 	line_draw();
 }