lineedit: first shot at optional unicode bidi input support

function old new delta read_line_input 4886 5003 +117 in_uint16_table - 97 +97 in_interval_table - 78 +78 static.rtl_b - 68 +68 unicode_isrtl - 55 +55 isrtl_str - 51 +51 static.rtl_p - 42 +42 unicode_conv_to_printable2 633 477 -156 ------------------------------------------------------------------------------ (add/remove: 6/0 grow/shrink: 1/1 up/down: 508/-156) Total: 352 bytes Signed-off-by: Tomas Heinrich <heinrich.tomas@gmail.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
author: Tomas Heinrich <heinrich.tomas@gmail.com> 2010-03-18 18:35:37 +0100
committer: Denys Vlasenko <vda.linux@googlemail.com> 2010-03-18 18:35:37 +0100
commit: c5c006c10c060e7f1a97250d039051b93ed390b2 (patch)
tree: b281136c99e6a27a530282a0b1b7eaf938704bb2
parent: 98f1dc12f1554aca6c3743bec1c3d8982a077f7c (diff)
download: busybox-w32-c5c006c10c060e7f1a97250d039051b93ed390b2.tar.gz
busybox-w32-c5c006c10c060e7f1a97250d039051b93ed390b2.tar.bz2
busybox-w32-c5c006c10c060e7f1a97250d039051b93ed390b2.zip
5 files changed, 189 insertions, 18 deletions
diff --git a/Config.in b/Config.in
index e7bb05dce..e0c01f3ef 100644
--- a/Config.in
+++ b/Config.in
@@ -196,6 +196,14 @@ config UNICODE_WIDE_WCHARS
          With this option off, any Unicode char with width > 1
          is substituted on output.
+config UNICODE_BIDI_SUPPORT
+        bool "Bidirectional character-aware line input"
+        default y
+        depends on FEATURE_ASSUME_UNICODE && !LOCALE_SUPPORT
+        help
+          With this option on, right-to-left Unicode characters
+          are treated differently on input (e.g. cursor movement).
 config LONG_OPTS
        bool "Support for --long-options"
        default y
diff --git a/include/unicode.h b/include/unicode.h
index 857aab138..05bdbca02 100644
--- a/include/unicode.h
+++ b/include/unicode.h
@@ -18,6 +18,8 @@ enum {
        UNICODE_ON = 2,
 };
+#define unicode_isrtl(wc) 0
 #if !ENABLE_FEATURE_ASSUME_UNICODE
 # define unicode_strlen(string) strlen(string)
@@ -26,6 +28,17 @@ enum {
 #else
+# if CONFIG_LAST_SUPPORTED_WCHAR < 126 || CONFIG_LAST_SUPPORTED_WCHAR >= 0x30000
+#  define LAST_SUPPORTED_WCHAR 0x2ffff
+# else
+#  define LAST_SUPPORTED_WCHAR CONFIG_LAST_SUPPORTED_WCHAR
+# endif
+# if LAST_SUPPORTED_WCHAR < 0x590
+#  undef  ENABLE_UNICODE_BIDI_SUPPORT
+#  define ENABLE_UNICODE_BIDI_SUPPORT 0
+# endif
 size_t FAST_FUNC unicode_strlen(const char *string);
 enum {
        UNI_FLAG_PAD = (1 << 0),
@@ -78,6 +91,10 @@ size_t wcrtomb(char *s, wchar_t wc, mbstate_t *ps) FAST_FUNC;
 int iswspace(wint_t wc) FAST_FUNC;
 int iswalnum(wint_t wc) FAST_FUNC;
 int iswpunct(wint_t wc) FAST_FUNC;
+#  if ENABLE_UNICODE_BIDI_SUPPORT
+#   undef unicode_isrtl
+int unicode_isrtl(wint_t wc) FAST_FUNC;
+#  endif
 # endif /* !LOCALE_SUPPORT */
diff --git a/libbb/lineedit.c b/libbb/lineedit.c
index 7c0eef90d..be022e8ae 100644
--- a/libbb/lineedit.c
+++ b/libbb/lineedit.c
@@ -1738,6 +1738,18 @@ static int lineedit_read_key(char *read_key_buffer)
        return ic;
 }
+#if ENABLE_UNICODE_BIDI_SUPPORT
+static int isrtl_str(void)
+{
+        int idx = cursor;
+        while (command_ps[idx] >= ' ' && command_ps[idx] < 127 && !isalpha(command_ps[idx]))
+                idx++;
+        return unicode_isrtl(command_ps[idx]);
+}
+#else
+# define isrtl_str() 0
+#endif
 /* leave out the "vi-mode"-only case labels if vi editing isn't
 * configured. */
 #define vi_case(caselabel) IF_FEATURE_EDITING_VI(case caselabel)
@@ -1895,10 +1907,9 @@ int FAST_FUNC read_line_input(const char *prompt, char *command, int maxsize, li
                        break;
                case CTRL('B'):
                vi_case('h'|VI_CMDMODE_BIT:)
-                vi_case('\b'|VI_CMDMODE_BIT:)
+                vi_case('\b'|VI_CMDMODE_BIT:) /* ^H */
                vi_case('\x7f'|VI_CMDMODE_BIT:) /* DEL */
-                        /* Control-b -- Move back one character */
+                        input_backward(1); /* Move back one character */
-                        input_backward(1);
                        break;
                case CTRL('E'):
                vi_case('$'|VI_CMDMODE_BIT:)
@@ -1908,13 +1919,20 @@ int FAST_FUNC read_line_input(const char *prompt, char *command, int maxsize, li
                case CTRL('F'):
                vi_case('l'|VI_CMDMODE_BIT:)
                vi_case(' '|VI_CMDMODE_BIT:)
-                        /* Control-f -- Move forward one character */
+                        input_forward(); /* Move forward one character */
-                        input_forward();
                        break;
-                case '\b':
+                case '\b':   /* ^H */
                case '\x7f': /* DEL */
-                        /* Control-h and DEL */
+                        if (!isrtl_str())
-                        input_backspace();
+                                input_backspace();
+                        else
+                                input_delete(0);
+                        break;
+                case KEYCODE_DELETE:
+                        if (!isrtl_str())
+                                input_delete(0);
+                        else
+                                input_backspace();
                        break;
 #if ENABLE_FEATURE_TAB_COMPLETION
                case '\t':
@@ -2137,9 +2155,6 @@ int FAST_FUNC read_line_input(const char *prompt, char *command, int maxsize, li
                case KEYCODE_CTRL_RIGHT:
                        ctrl_right();
                        break;
-                case KEYCODE_DELETE:
-                        input_delete(0);
-                        break;
                case KEYCODE_HOME:
                        input_backward(cursor);
                        break;
@@ -2205,14 +2220,19 @@ int FAST_FUNC read_line_input(const char *prompt, char *command, int maxsize, li
                                command_ps[cursor] = ic;
                                command_ps[cursor + 1] = BB_NUL;
                                cmdedit_set_out_char(' ');
+                                if (unicode_isrtl(ic))
+                                        input_backward(1);
                        } else {
                                /* In the middle, insert */
+                                /* is char right-to-left, or "neutral" one (e.g. comma) added to rtl text? */
+                                int rtl = ENABLE_UNICODE_BIDI_SUPPORT ? (unicode_isrtl(ic) || (ic < 127 && !isalpha(ic) && isrtl_str())) : 0;
                                int sc = cursor;
                                memmove(command_ps + sc + 1, command_ps + sc,
                                        (command_len - sc) * sizeof(command_ps[0]));
                                command_ps[sc] = ic;
-                                sc++;
+                                if (!rtl)
+                                        sc++;
                                /* rewrite from cursor */
                                input_end();
                                /* to prev x pos + 1 */
diff --git a/libbb/unicode.c b/libbb/unicode.c
index 7c41ef30b..91667ea72 100644
--- a/libbb/unicode.c
+++ b/libbb/unicode.c
@@ -241,6 +241,138 @@ int FAST_FUNC iswpunct(wint_t wc)
 #include "unicode_wcwidth.c"
+# if ENABLE_UNICODE_BIDI_SUPPORT
+int FAST_FUNC unicode_isrtl(wint_t wc)
+{
+        /* ranges taken from
+         * http://www.unicode.org/Public/5.2.0/ucd/extracted/DerivedBidiClass.txt
+         * Bidi_Class=Left_To_Right | Bidi_Class=Arabic_Letter
+         */
+        static const struct interval rtl_b[] = {
+#  define BIG_(a,b) { a, b },
+#  define PAIR(a,b)
+                PAIR(0x0590, 0x0590)
+                PAIR(0x05BE, 0x05BE)
+                PAIR(0x05C0, 0x05C0)
+                PAIR(0x05C3, 0x05C3)
+                PAIR(0x05C6, 0x05C6)
+                BIG_(0x05C8, 0x05FF)
+                PAIR(0x0604, 0x0605)
+                PAIR(0x0608, 0x0608)
+                PAIR(0x060B, 0x060B)
+                PAIR(0x060D, 0x060D)
+                BIG_(0x061B, 0x064A)
+                PAIR(0x065F, 0x065F)
+                PAIR(0x066D, 0x066F)
+                BIG_(0x0671, 0x06D5)
+                PAIR(0x06E5, 0x06E6)
+                PAIR(0x06EE, 0x06EF)
+                BIG_(0x06FA, 0x070E)
+                PAIR(0x0710, 0x0710)
+                BIG_(0x0712, 0x072F)
+                BIG_(0x074B, 0x07A5)
+                BIG_(0x07B1, 0x07EA)
+                PAIR(0x07F4, 0x07F5)
+                BIG_(0x07FA, 0x0815)
+                PAIR(0x081A, 0x081A)
+                PAIR(0x0824, 0x0824)
+                PAIR(0x0828, 0x0828)
+                BIG_(0x082E, 0x08FF)
+                PAIR(0x200F, 0x200F)
+                PAIR(0x202B, 0x202B)
+                PAIR(0x202E, 0x202E)
+                BIG_(0xFB1D, 0xFB1D)
+                BIG_(0xFB1F, 0xFB28)
+                BIG_(0xFB2A, 0xFD3D)
+                BIG_(0xFD40, 0xFDCF)
+                BIG_(0xFDC8, 0xFDCF)
+                BIG_(0xFDF0, 0xFDFC)
+                BIG_(0xFDFE, 0xFDFF)
+                BIG_(0xFE70, 0xFEFE)
+                /* Probably not necessary
+                {0x10800, 0x1091E},
+                {0x10920, 0x10A00},
+                {0x10A04, 0x10A04},
+                {0x10A07, 0x10A0B},
+                {0x10A10, 0x10A37},
+                {0x10A3B, 0x10A3E},
+                {0x10A40, 0x10A7F},
+                {0x10B36, 0x10B38},
+                {0x10B40, 0x10E5F},
+                {0x10E7F, 0x10FFF},
+                {0x1E800, 0x1EFFF}
+                */
+#  undef BIG_
+#  undef PAIR
+        };
+        static const uint16_t rtl_p[] = {
+#  define BIG_(a,b)
+#  define PAIR(a,b) (a << 2) | (b-a),
+                /* Exact copy-n-paste of the above: */
+                PAIR(0x0590, 0x0590)
+                PAIR(0x05BE, 0x05BE)
+                PAIR(0x05C0, 0x05C0)
+                PAIR(0x05C3, 0x05C3)
+                PAIR(0x05C6, 0x05C6)
+                BIG_(0x05C8, 0x05FF)
+                PAIR(0x0604, 0x0605)
+                PAIR(0x0608, 0x0608)
+                PAIR(0x060B, 0x060B)
+                PAIR(0x060D, 0x060D)
+                BIG_(0x061B, 0x064A)
+                PAIR(0x065F, 0x065F)
+                PAIR(0x066D, 0x066F)
+                BIG_(0x0671, 0x06D5)
+                PAIR(0x06E5, 0x06E6)
+                PAIR(0x06EE, 0x06EF)
+                BIG_(0x06FA, 0x070E)
+                PAIR(0x0710, 0x0710)
+                BIG_(0x0712, 0x072F)
+                BIG_(0x074B, 0x07A5)
+                BIG_(0x07B1, 0x07EA)
+                PAIR(0x07F4, 0x07F5)
+                BIG_(0x07FA, 0x0815)
+                PAIR(0x081A, 0x081A)
+                PAIR(0x0824, 0x0824)
+                PAIR(0x0828, 0x0828)
+                BIG_(0x082E, 0x08FF)
+                PAIR(0x200F, 0x200F)
+                PAIR(0x202B, 0x202B)
+                PAIR(0x202E, 0x202E)
+                BIG_(0xFB1D, 0xFB1D)
+                BIG_(0xFB1F, 0xFB28)
+                BIG_(0xFB2A, 0xFD3D)
+                BIG_(0xFD40, 0xFDCF)
+                BIG_(0xFDC8, 0xFDCF)
+                BIG_(0xFDF0, 0xFDFC)
+                BIG_(0xFDFE, 0xFDFF)
+                BIG_(0xFE70, 0xFEFE)
+                /* Probably not necessary
+                {0x10800, 0x1091E},
+                {0x10920, 0x10A00},
+                {0x10A04, 0x10A04},
+                {0x10A07, 0x10A0B},
+                {0x10A10, 0x10A37},
+                {0x10A3B, 0x10A3E},
+                {0x10A40, 0x10A7F},
+                {0x10B36, 0x10B38},
+                {0x10B40, 0x10E5F},
+                {0x10E7F, 0x10FFF},
+                {0x1E800, 0x1EFFF}
+                */
+#  undef BIG_
+#  undef PAIR
+        };
+        if (in_interval_table(wc, rtl_b, ARRAY_SIZE(rtl_b) - 1))
+                return 1;
+        if (in_uint16_table(wc, rtl_p, ARRAY_SIZE(rtl_p) - 1))
+                return 1;
+        return 0;
+}
+# endif /* UNICODE_BIDI_SUPPORT */
 #endif /* Homegrown Unicode support */
diff --git a/libbb/unicode_wcwidth.c b/libbb/unicode_wcwidth.c
index a81a98038..7eccc394c 100644
--- a/libbb/unicode_wcwidth.c
+++ b/libbb/unicode_wcwidth.c
@@ -90,12 +90,6 @@
 * until Unicode committee assigns something there.
 */
-#if CONFIG_LAST_SUPPORTED_WCHAR < 126 || CONFIG_LAST_SUPPORTED_WCHAR >= 0x30000
-# define LAST_SUPPORTED_WCHAR 0x2ffff
-#else
-# define LAST_SUPPORTED_WCHAR CONFIG_LAST_SUPPORTED_WCHAR
-#endif
 #if LAST_SUPPORTED_WCHAR >= 0x300
 struct interval {
        uint16_t first;
author	Tomas Heinrich <heinrich.tomas@gmail.com>	2010-03-18 18:35:37 +0100
committer	Denys Vlasenko <vda.linux@googlemail.com>	2010-03-18 18:35:37 +0100
commit	c5c006c10c060e7f1a97250d039051b93ed390b2 (patch)
tree	b281136c99e6a27a530282a0b1b7eaf938704bb2
parent	98f1dc12f1554aca6c3743bec1c3d8982a077f7c (diff)
download	busybox-w32-c5c006c10c060e7f1a97250d039051b93ed390b2.tar.gz busybox-w32-c5c006c10c060e7f1a97250d039051b93ed390b2.tar.bz2 busybox-w32-c5c006c10c060e7f1a97250d039051b93ed390b2.zip

diff --git a/Config.in b/Config.in index e7bb05dce..e0c01f3ef 100644 --- a/Config.in +++ b/Config.in
@@ -196,6 +196,14 @@ config UNICODE_WIDE_WCHARS
196	With this option off, any Unicode char with width > 1	196	With this option off, any Unicode char with width > 1
197	is substituted on output.	197	is substituted on output.
198		198
		199	config UNICODE_BIDI_SUPPORT
		200	bool "Bidirectional character-aware line input"
		201	default y
		202	depends on FEATURE_ASSUME_UNICODE && !LOCALE_SUPPORT
		203	help
		204	With this option on, right-to-left Unicode characters
		205	are treated differently on input (e.g. cursor movement).
		206
199	config LONG_OPTS	207	config LONG_OPTS
200	bool "Support for --long-options"	208	bool "Support for --long-options"
201	default y	209	default y


diff --git a/include/unicode.h b/include/unicode.h index 857aab138..05bdbca02 100644 --- a/include/unicode.h +++ b/include/unicode.h
@@ -18,6 +18,8 @@ enum {
18	UNICODE_ON = 2,	18	UNICODE_ON = 2,
19	};	19	};
20		20
		21	#define unicode_isrtl(wc) 0
		22
21	#if !ENABLE_FEATURE_ASSUME_UNICODE	23	#if !ENABLE_FEATURE_ASSUME_UNICODE
22		24
23	# define unicode_strlen(string) strlen(string)	25	# define unicode_strlen(string) strlen(string)
@@ -26,6 +28,17 @@ enum {
26		28
27	#else	29	#else
28		30
		31	# if CONFIG_LAST_SUPPORTED_WCHAR < 126 \|\| CONFIG_LAST_SUPPORTED_WCHAR >= 0x30000
		32	# define LAST_SUPPORTED_WCHAR 0x2ffff
		33	# else
		34	# define LAST_SUPPORTED_WCHAR CONFIG_LAST_SUPPORTED_WCHAR
		35	# endif
		36
		37	# if LAST_SUPPORTED_WCHAR < 0x590
		38	# undef ENABLE_UNICODE_BIDI_SUPPORT
		39	# define ENABLE_UNICODE_BIDI_SUPPORT 0
		40	# endif
		41
29	size_t FAST_FUNC unicode_strlen(const char *string);	42	size_t FAST_FUNC unicode_strlen(const char *string);
30	enum {	43	enum {
31	UNI_FLAG_PAD = (1 << 0),	44	UNI_FLAG_PAD = (1 << 0),
@@ -78,6 +91,10 @@ size_t wcrtomb(char s, wchar_t wc, mbstate_t ps) FAST_FUNC;
78	int iswspace(wint_t wc) FAST_FUNC;	91	int iswspace(wint_t wc) FAST_FUNC;
79	int iswalnum(wint_t wc) FAST_FUNC;	92	int iswalnum(wint_t wc) FAST_FUNC;
80	int iswpunct(wint_t wc) FAST_FUNC;	93	int iswpunct(wint_t wc) FAST_FUNC;
		94	# if ENABLE_UNICODE_BIDI_SUPPORT
		95	# undef unicode_isrtl
		96	int unicode_isrtl(wint_t wc) FAST_FUNC;
		97	# endif
81		98
82		99
83	# endif /* !LOCALE_SUPPORT */	100	# endif /* !LOCALE_SUPPORT */


diff --git a/libbb/lineedit.c b/libbb/lineedit.c index 7c0eef90d..be022e8ae 100644 --- a/libbb/lineedit.c +++ b/libbb/lineedit.c
@@ -1738,6 +1738,18 @@ static int lineedit_read_key(char *read_key_buffer)
1738	return ic;	1738	return ic;
1739	}	1739	}
1740		1740
		1741	#if ENABLE_UNICODE_BIDI_SUPPORT
		1742	static int isrtl_str(void)
		1743	{
		1744	int idx = cursor;
		1745	while (command_ps[idx] >= ' ' && command_ps[idx] < 127 && !isalpha(command_ps[idx]))
		1746	idx++;
		1747	return unicode_isrtl(command_ps[idx]);
		1748	}
		1749	#else
		1750	# define isrtl_str() 0
		1751	#endif
		1752
1741	/* leave out the "vi-mode"-only case labels if vi editing isn't	1753	/* leave out the "vi-mode"-only case labels if vi editing isn't
1742	* configured. */	1754	* configured. */
1743	#define vi_case(caselabel) IF_FEATURE_EDITING_VI(case caselabel)	1755	#define vi_case(caselabel) IF_FEATURE_EDITING_VI(case caselabel)
@@ -1895,10 +1907,9 @@ int FAST_FUNC read_line_input(const char prompt, char command, int maxsize, li
1895	break;	1907	break;
1896	case CTRL('B'):	1908	case CTRL('B'):
1897	vi_case('h'\|VI_CMDMODE_BIT:)	1909	vi_case('h'\|VI_CMDMODE_BIT:)
1898	vi_case('\b'\|VI_CMDMODE_BIT:)	1910	vi_case('\b'\|VI_CMDMODE_BIT:) /* ^H */
1899	vi_case('\x7f'\|VI_CMDMODE_BIT:) /* DEL */	1911	vi_case('\x7f'\|VI_CMDMODE_BIT:) /* DEL */
1900	/* Control-b -- Move back one character */	1912	input_backward(1); /* Move back one character */
1901	input_backward(1);
1902	break;	1913	break;
1903	case CTRL('E'):	1914	case CTRL('E'):
1904	vi_case('$'\|VI_CMDMODE_BIT:)	1915	vi_case('$'\|VI_CMDMODE_BIT:)
@@ -1908,13 +1919,20 @@ int FAST_FUNC read_line_input(const char prompt, char command, int maxsize, li
1908	case CTRL('F'):	1919	case CTRL('F'):
1909	vi_case('l'\|VI_CMDMODE_BIT:)	1920	vi_case('l'\|VI_CMDMODE_BIT:)
1910	vi_case(' '\|VI_CMDMODE_BIT:)	1921	vi_case(' '\|VI_CMDMODE_BIT:)
1911	/* Control-f -- Move forward one character */	1922	input_forward(); /* Move forward one character */
1912	input_forward();
1913	break;	1923	break;
1914	case '\b':	1924	case '\b': /* ^H */
1915	case '\x7f': /* DEL */	1925	case '\x7f': /* DEL */
1916	/* Control-h and DEL */	1926	if (!isrtl_str())
1917	input_backspace();	1927	input_backspace();
		1928	else
		1929	input_delete(0);
		1930	break;
		1931	case KEYCODE_DELETE:
		1932	if (!isrtl_str())
		1933	input_delete(0);
		1934	else
		1935	input_backspace();
1918	break;	1936	break;
1919	#if ENABLE_FEATURE_TAB_COMPLETION	1937	#if ENABLE_FEATURE_TAB_COMPLETION
1920	case '\t':	1938	case '\t':
@@ -2137,9 +2155,6 @@ int FAST_FUNC read_line_input(const char prompt, char command, int maxsize, li
2137	case KEYCODE_CTRL_RIGHT:	2155	case KEYCODE_CTRL_RIGHT:
2138	ctrl_right();	2156	ctrl_right();
2139	break;	2157	break;
2140	case KEYCODE_DELETE:
2141	input_delete(0);
2142	break;
2143	case KEYCODE_HOME:	2158	case KEYCODE_HOME:
2144	input_backward(cursor);	2159	input_backward(cursor);
2145	break;	2160	break;
@@ -2205,14 +2220,19 @@ int FAST_FUNC read_line_input(const char prompt, char command, int maxsize, li
2205	command_ps[cursor] = ic;	2220	command_ps[cursor] = ic;
2206	command_ps[cursor + 1] = BB_NUL;	2221	command_ps[cursor + 1] = BB_NUL;
2207	cmdedit_set_out_char(' ');	2222	cmdedit_set_out_char(' ');
		2223	if (unicode_isrtl(ic))
		2224	input_backward(1);
2208	} else {	2225	} else {
2209	/* In the middle, insert */	2226	/* In the middle, insert */
		2227	/* is char right-to-left, or "neutral" one (e.g. comma) added to rtl text? */
		2228	int rtl = ENABLE_UNICODE_BIDI_SUPPORT ? (unicode_isrtl(ic) \|\| (ic < 127 && !isalpha(ic) && isrtl_str())) : 0;
2210	int sc = cursor;	2229	int sc = cursor;
2211		2230
2212	memmove(command_ps + sc + 1, command_ps + sc,	2231	memmove(command_ps + sc + 1, command_ps + sc,
2213	(command_len - sc) * sizeof(command_ps[0]));	2232	(command_len - sc) * sizeof(command_ps[0]));
2214	command_ps[sc] = ic;	2233	command_ps[sc] = ic;
2215	sc++;	2234	if (!rtl)
		2235	sc++;
2216	/* rewrite from cursor */	2236	/* rewrite from cursor */
2217	input_end();	2237	input_end();
2218	/* to prev x pos + 1 */	2238	/* to prev x pos + 1 */


diff --git a/libbb/unicode.c b/libbb/unicode.c index 7c41ef30b..91667ea72 100644 --- a/libbb/unicode.c +++ b/libbb/unicode.c
@@ -241,6 +241,138 @@ int FAST_FUNC iswpunct(wint_t wc)
241		241
242	#include "unicode_wcwidth.c"	242	#include "unicode_wcwidth.c"
243		243
		244	# if ENABLE_UNICODE_BIDI_SUPPORT
		245	int FAST_FUNC unicode_isrtl(wint_t wc)
		246	{
		247	/* ranges taken from
		248	* http://www.unicode.org/Public/5.2.0/ucd/extracted/DerivedBidiClass.txt
		249	* Bidi_Class=Left_To_Right \| Bidi_Class=Arabic_Letter
		250	*/
		251	static const struct interval rtl_b[] = {
		252	# define BIG_(a,b) { a, b },
		253	# define PAIR(a,b)
		254	PAIR(0x0590, 0x0590)
		255	PAIR(0x05BE, 0x05BE)
		256	PAIR(0x05C0, 0x05C0)
		257	PAIR(0x05C3, 0x05C3)
		258	PAIR(0x05C6, 0x05C6)
		259	BIG_(0x05C8, 0x05FF)
		260	PAIR(0x0604, 0x0605)
		261	PAIR(0x0608, 0x0608)
		262	PAIR(0x060B, 0x060B)
		263	PAIR(0x060D, 0x060D)
		264	BIG_(0x061B, 0x064A)
		265	PAIR(0x065F, 0x065F)
		266	PAIR(0x066D, 0x066F)
		267	BIG_(0x0671, 0x06D5)
		268	PAIR(0x06E5, 0x06E6)
		269	PAIR(0x06EE, 0x06EF)
		270	BIG_(0x06FA, 0x070E)
		271	PAIR(0x0710, 0x0710)
		272	BIG_(0x0712, 0x072F)
		273	BIG_(0x074B, 0x07A5)
		274	BIG_(0x07B1, 0x07EA)
		275	PAIR(0x07F4, 0x07F5)
		276	BIG_(0x07FA, 0x0815)
		277	PAIR(0x081A, 0x081A)
		278	PAIR(0x0824, 0x0824)
		279	PAIR(0x0828, 0x0828)
		280	BIG_(0x082E, 0x08FF)
		281	PAIR(0x200F, 0x200F)
		282	PAIR(0x202B, 0x202B)
		283	PAIR(0x202E, 0x202E)
		284	BIG_(0xFB1D, 0xFB1D)
		285	BIG_(0xFB1F, 0xFB28)
		286	BIG_(0xFB2A, 0xFD3D)
		287	BIG_(0xFD40, 0xFDCF)
		288	BIG_(0xFDC8, 0xFDCF)
		289	BIG_(0xFDF0, 0xFDFC)
		290	BIG_(0xFDFE, 0xFDFF)
		291	BIG_(0xFE70, 0xFEFE)
		292	/* Probably not necessary
		293	{0x10800, 0x1091E},
		294	{0x10920, 0x10A00},
		295	{0x10A04, 0x10A04},
		296	{0x10A07, 0x10A0B},
		297	{0x10A10, 0x10A37},
		298	{0x10A3B, 0x10A3E},
		299	{0x10A40, 0x10A7F},
		300	{0x10B36, 0x10B38},
		301	{0x10B40, 0x10E5F},
		302	{0x10E7F, 0x10FFF},
		303	{0x1E800, 0x1EFFF}
		304	*/
		305	# undef BIG_
		306	# undef PAIR
		307	};
		308
		309	static const uint16_t rtl_p[] = {
		310	# define BIG_(a,b)
		311	# define PAIR(a,b) (a << 2) \| (b-a),
		312	/* Exact copy-n-paste of the above: */
		313	PAIR(0x0590, 0x0590)
		314	PAIR(0x05BE, 0x05BE)
		315	PAIR(0x05C0, 0x05C0)
		316	PAIR(0x05C3, 0x05C3)
		317	PAIR(0x05C6, 0x05C6)
		318	BIG_(0x05C8, 0x05FF)
		319	PAIR(0x0604, 0x0605)
		320	PAIR(0x0608, 0x0608)
		321	PAIR(0x060B, 0x060B)
		322	PAIR(0x060D, 0x060D)
		323	BIG_(0x061B, 0x064A)
		324	PAIR(0x065F, 0x065F)
		325	PAIR(0x066D, 0x066F)
		326	BIG_(0x0671, 0x06D5)
		327	PAIR(0x06E5, 0x06E6)
		328	PAIR(0x06EE, 0x06EF)
		329	BIG_(0x06FA, 0x070E)
		330	PAIR(0x0710, 0x0710)
		331	BIG_(0x0712, 0x072F)
		332	BIG_(0x074B, 0x07A5)
		333	BIG_(0x07B1, 0x07EA)
		334	PAIR(0x07F4, 0x07F5)
		335	BIG_(0x07FA, 0x0815)
		336	PAIR(0x081A, 0x081A)
		337	PAIR(0x0824, 0x0824)
		338	PAIR(0x0828, 0x0828)
		339	BIG_(0x082E, 0x08FF)
		340	PAIR(0x200F, 0x200F)
		341	PAIR(0x202B, 0x202B)
		342	PAIR(0x202E, 0x202E)
		343	BIG_(0xFB1D, 0xFB1D)
		344	BIG_(0xFB1F, 0xFB28)
		345	BIG_(0xFB2A, 0xFD3D)
		346	BIG_(0xFD40, 0xFDCF)
		347	BIG_(0xFDC8, 0xFDCF)
		348	BIG_(0xFDF0, 0xFDFC)
		349	BIG_(0xFDFE, 0xFDFF)
		350	BIG_(0xFE70, 0xFEFE)
		351	/* Probably not necessary
		352	{0x10800, 0x1091E},
		353	{0x10920, 0x10A00},
		354	{0x10A04, 0x10A04},
		355	{0x10A07, 0x10A0B},
		356	{0x10A10, 0x10A37},
		357	{0x10A3B, 0x10A3E},
		358	{0x10A40, 0x10A7F},
		359	{0x10B36, 0x10B38},
		360	{0x10B40, 0x10E5F},
		361	{0x10E7F, 0x10FFF},
		362	{0x1E800, 0x1EFFF}
		363	*/
		364	# undef BIG_
		365	# undef PAIR
		366	};
		367
		368	if (in_interval_table(wc, rtl_b, ARRAY_SIZE(rtl_b) - 1))
		369	return 1;
		370	if (in_uint16_table(wc, rtl_p, ARRAY_SIZE(rtl_p) - 1))
		371	return 1;
		372	return 0;
		373	}
		374	# endif /* UNICODE_BIDI_SUPPORT */
		375
244	#endif /* Homegrown Unicode support */	376	#endif /* Homegrown Unicode support */
245		377
246		378


diff --git a/libbb/unicode_wcwidth.c b/libbb/unicode_wcwidth.c index a81a98038..7eccc394c 100644 --- a/libbb/unicode_wcwidth.c +++ b/libbb/unicode_wcwidth.c
@@ -90,12 +90,6 @@
90	* until Unicode committee assigns something there.	90	* until Unicode committee assigns something there.
91	*/	91	*/
92		92
93	#if CONFIG_LAST_SUPPORTED_WCHAR < 126 \|\| CONFIG_LAST_SUPPORTED_WCHAR >= 0x30000
94	# define LAST_SUPPORTED_WCHAR 0x2ffff
95	#else
96	# define LAST_SUPPORTED_WCHAR CONFIG_LAST_SUPPORTED_WCHAR
97	#endif
98
99	#if LAST_SUPPORTED_WCHAR >= 0x300	93	#if LAST_SUPPORTED_WCHAR >= 0x300
100	struct interval {	94	struct interval {
101	uint16_t first;	95	uint16_t first;