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>

5 files changed, 189 insertions, 18 deletions

diff --git a/Config.in b/Config.in
index e7bb05d..e0c01f3 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 857aab1..05bdbca 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 7c0eef9..be022e8 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);
+			input_backward(1); /* Move back one character */
 			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();
+			input_forward(); /* Move forward one character */
 			break;
-		case '\b':
+		case '\b':   /* ^H */
 		case '\x7f': /* DEL */
-			/* Control-h and DEL */
-			input_backspace();
+			if (!isrtl_str())
+				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 7c41ef3..91667ea 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 a81a980..7eccc39 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;