unicode: optional table for better handling of neutral bidi chars

Off:
function                                             old     new   delta
unicode_bidi_isrtl                                     -      55     +55
isrtl_str                                             51      65     +14
unicode_isrtl                                         55       -     -55
read_line_input                                     5003    4937     -66
------------------------------------------------------------------------------
(add/remove: 1/4 grow/shrink: 1/1 up/down: 69/-121)           Total: -52 bytes

On:
function                                             old     new   delta
static.neutral_b                                       -     320    +320
static.neutral_p                                       -     142    +142
unicode_bidi_isrtl                                     -      55     +55
unicode_bidi_is_neutral_wchar                          -      55     +55
isrtl_str                                             51      59      +8
unicode_isrtl                                         55       -     -55
read_line_input                                     5003    4937     -66
------------------------------------------------------------------------------
(add/remove: 4/4 grow/shrink: 1/1 up/down: 580/-121)          Total: 459 bytes

Signed-off-by: Tomas Heinrich <heinrich.tomas@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

5 files changed, 284 insertions, 106 deletions

diff --git a/Config.in b/Config.in
index e0c01f3ef..4439ce4f9 100644
--- a/Config.in
+++ b/Config.in
@@ -198,12 +198,23 @@ config UNICODE_WIDE_WCHARS
 
 config UNICODE_BIDI_SUPPORT
 	bool "Bidirectional character-aware line input"
-	default y
+	default n
 	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 UNICODE_NEUTRAL_TABLE
+	bool "In bidi input, support non-ASCII neutral chars too"
+	default n
+	depends on UNICODE_BIDI_SUPPORT
+	help
+	  In most cases it's enough to treat only ASCII non-letters
+	  (i.e. punctuation, numbers and space) as characters
+	  with neutral directionality.
+	  With this option on, more extensive (and bigger) table
+	  of neutral chars will be used.
+
 config LONG_OPTS
 	bool "Support for --long-options"
 	default y
diff --git a/include/unicode.h b/include/unicode.h
index 05bdbca02..deb4022c3 100644
--- a/include/unicode.h
+++ b/include/unicode.h
@@ -18,7 +18,8 @@ enum {
 	UNICODE_ON = 2,
 };
 
-#define unicode_isrtl(wc) 0
+#define unicode_bidi_isrtl(wc) 0
+#define unicode_bidi_is_neutral_wchar(wc) (wc <= 126 && !isalpha(wc))
 
 #if !ENABLE_FEATURE_ASSUME_UNICODE
 
@@ -92,8 +93,12 @@ 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;
+#   undef unicode_bidi_isrtl
+int unicode_bidi_isrtl(wint_t wc) FAST_FUNC;
+#   if ENABLE_UNICODE_NEUTRAL_TABLE
+#    undef unicode_bidi_is_neutral_wchar
+int unicode_bidi_is_neutral_wchar(wint_t wc) FAST_FUNC;
+#   endif
 #  endif
 
 
diff --git a/libbb/lineedit.c b/libbb/lineedit.c
index be022e8ae..38a09cb26 100644
--- a/libbb/lineedit.c
+++ b/libbb/lineedit.c
@@ -1742,9 +1742,10 @@ static int lineedit_read_key(char *read_key_buffer)
 static int isrtl_str(void)
 {
 	int idx = cursor;
-	while (command_ps[idx] >= ' ' && command_ps[idx] < 127 && !isalpha(command_ps[idx]))
+
+	while (idx < command_len && unicode_bidi_is_neutral_wchar(command_ps[idx]))
 		idx++;
-	return unicode_isrtl(command_ps[idx]);
+	return unicode_bidi_isrtl(command_ps[idx]);
 }
 #else
 # define isrtl_str() 0
@@ -2220,19 +2221,18 @@ 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))
+				if (unicode_bidi_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;
-				if (!rtl)
-					sc++;
+				/* is right-to-left char, or neutral one (e.g. comma) was just added to rtl text? */
+				if (!isrtl_str())
+					sc++; /* no */
 				/* rewrite from cursor */
 				input_end();
 				/* to prev x pos + 1 */
diff --git a/libbb/unicode.c b/libbb/unicode.c
index 91667ea72..bc9714562 100644
--- a/libbb/unicode.c
+++ b/libbb/unicode.c
@@ -242,7 +242,7 @@ int FAST_FUNC iswpunct(wint_t wc)
 #include "unicode_wcwidth.c"
 
 # if ENABLE_UNICODE_BIDI_SUPPORT
-int FAST_FUNC unicode_isrtl(wint_t wc)
+int FAST_FUNC unicode_bidi_isrtl(wint_t wc)
 {
 	/* ranges taken from
 	 * http://www.unicode.org/Public/5.2.0/ucd/extracted/DerivedBidiClass.txt
@@ -251,43 +251,44 @@ int FAST_FUNC unicode_isrtl(wint_t wc)
 	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)
+#  define ARRAY \
+		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},
@@ -302,68 +303,21 @@ int FAST_FUNC unicode_isrtl(wint_t wc)
 		{0x10E7F, 0x10FFF},
 		{0x1E800, 0x1EFFF}
 		*/
+		ARRAY
 #  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}
-		*/
+	static const uint16_t rtl_p[] = { ARRAY };
 #  undef BIG_
 #  undef PAIR
-	};
+#  define BIG_(a,b) char big_##a[b < 0x4000 && b-a <= 3 ? -1 : 1];
+#  define PAIR(a,b) char pair##a[b >= 0x4000 || b-a > 3 ? -1 : 1];
+	struct CHECK { ARRAY };
+#  undef BIG_
+#  undef PAIR
+#  undef ARRAY
 
 	if (in_interval_table(wc, rtl_b, ARRAY_SIZE(rtl_b) - 1))
 		return 1;
@@ -371,6 +325,214 @@ int FAST_FUNC unicode_isrtl(wint_t wc)
 		return 1;
 	return 0;
 }
+
+#  if ENABLE_UNICODE_NEUTRAL_TABLE
+int FAST_FUNC unicode_bidi_is_neutral_wchar(wint_t wc)
+{
+	/* ranges taken from
+	 * http://www.unicode.org/Public/5.2.0/ucd/extracted/DerivedBidiClass.txt
+	 * Bidi_Classes: Paragraph_Separator, Segment_Separator,
+	 * White_Space, Other_Neutral, European_Number, European_Separator,
+	 * European_Terminator, Arabic_Number, Common_Separator
+	 */
+	static const struct interval neutral_b[] = {
+#  define BIG_(a,b) { a, b },
+#  define PAIR(a,b)
+#  define ARRAY \
+		BIG_(0x0009, 0x000D) \
+		BIG_(0x001C, 0x0040) \
+		BIG_(0x005B, 0x0060) \
+		PAIR(0x007B, 0x007E) \
+		PAIR(0x0085, 0x0085) \
+		BIG_(0x00A0, 0x00A9) \
+		PAIR(0x00AB, 0x00AC) \
+		BIG_(0x00AE, 0x00B4) \
+		PAIR(0x00B6, 0x00B9) \
+		BIG_(0x00BB, 0x00BF) \
+		PAIR(0x00D7, 0x00D7) \
+		PAIR(0x00F7, 0x00F7) \
+		PAIR(0x02B9, 0x02BA) \
+		BIG_(0x02C2, 0x02CF) \
+		BIG_(0x02D2, 0x02DF) \
+		BIG_(0x02E5, 0x02FF) \
+		PAIR(0x0374, 0x0375) \
+		PAIR(0x037E, 0x037E) \
+		PAIR(0x0384, 0x0385) \
+		PAIR(0x0387, 0x0387) \
+		PAIR(0x03F6, 0x03F6) \
+		PAIR(0x058A, 0x058A) \
+		PAIR(0x0600, 0x0603) \
+		PAIR(0x0606, 0x0607) \
+		PAIR(0x0609, 0x060A) \
+		PAIR(0x060C, 0x060C) \
+		PAIR(0x060E, 0x060F) \
+		BIG_(0x0660, 0x066C) \
+		PAIR(0x06DD, 0x06DD) \
+		PAIR(0x06E9, 0x06E9) \
+		BIG_(0x06F0, 0x06F9) \
+		PAIR(0x07F6, 0x07F9) \
+		PAIR(0x09F2, 0x09F3) \
+		PAIR(0x09FB, 0x09FB) \
+		PAIR(0x0AF1, 0x0AF1) \
+		BIG_(0x0BF3, 0x0BFA) \
+		BIG_(0x0C78, 0x0C7E) \
+		PAIR(0x0CF1, 0x0CF2) \
+		PAIR(0x0E3F, 0x0E3F) \
+		PAIR(0x0F3A, 0x0F3D) \
+		BIG_(0x1390, 0x1400) \
+		PAIR(0x1680, 0x1680) \
+		PAIR(0x169B, 0x169C) \
+		PAIR(0x17DB, 0x17DB) \
+		BIG_(0x17F0, 0x17F9) \
+		BIG_(0x1800, 0x180A) \
+		PAIR(0x180E, 0x180E) \
+		PAIR(0x1940, 0x1940) \
+		PAIR(0x1944, 0x1945) \
+		BIG_(0x19DE, 0x19FF) \
+		PAIR(0x1FBD, 0x1FBD) \
+		PAIR(0x1FBF, 0x1FC1) \
+		PAIR(0x1FCD, 0x1FCF) \
+		PAIR(0x1FDD, 0x1FDF) \
+		PAIR(0x1FED, 0x1FEF) \
+		PAIR(0x1FFD, 0x1FFE) \
+		BIG_(0x2000, 0x200A) \
+		BIG_(0x2010, 0x2029) \
+		BIG_(0x202F, 0x205F) \
+		PAIR(0x2070, 0x2070) \
+		BIG_(0x2074, 0x207E) \
+		BIG_(0x2080, 0x208E) \
+		BIG_(0x20A0, 0x20B8) \
+		PAIR(0x2100, 0x2101) \
+		PAIR(0x2103, 0x2106) \
+		PAIR(0x2108, 0x2109) \
+		PAIR(0x2114, 0x2114) \
+		PAIR(0x2116, 0x2118) \
+		BIG_(0x211E, 0x2123) \
+		PAIR(0x2125, 0x2125) \
+		PAIR(0x2127, 0x2127) \
+		PAIR(0x2129, 0x2129) \
+		PAIR(0x212E, 0x212E) \
+		PAIR(0x213A, 0x213B) \
+		BIG_(0x2140, 0x2144) \
+		PAIR(0x214A, 0x214D) \
+		BIG_(0x2150, 0x215F) \
+		PAIR(0x2189, 0x2189) \
+		BIG_(0x2190, 0x2335) \
+		BIG_(0x237B, 0x2394) \
+		BIG_(0x2396, 0x23E8) \
+		BIG_(0x2400, 0x2426) \
+		BIG_(0x2440, 0x244A) \
+		BIG_(0x2460, 0x249B) \
+		BIG_(0x24EA, 0x26AB) \
+		BIG_(0x26AD, 0x26CD) \
+		BIG_(0x26CF, 0x26E1) \
+		PAIR(0x26E3, 0x26E3) \
+		BIG_(0x26E8, 0x26FF) \
+		PAIR(0x2701, 0x2704) \
+		PAIR(0x2706, 0x2709) \
+		BIG_(0x270C, 0x2727) \
+		BIG_(0x2729, 0x274B) \
+		PAIR(0x274D, 0x274D) \
+		PAIR(0x274F, 0x2752) \
+		BIG_(0x2756, 0x275E) \
+		BIG_(0x2761, 0x2794) \
+		BIG_(0x2798, 0x27AF) \
+		BIG_(0x27B1, 0x27BE) \
+		BIG_(0x27C0, 0x27CA) \
+		PAIR(0x27CC, 0x27CC) \
+		BIG_(0x27D0, 0x27FF) \
+		BIG_(0x2900, 0x2B4C) \
+		BIG_(0x2B50, 0x2B59) \
+		BIG_(0x2CE5, 0x2CEA) \
+		BIG_(0x2CF9, 0x2CFF) \
+		BIG_(0x2E00, 0x2E99) \
+		BIG_(0x2E9B, 0x2EF3) \
+		BIG_(0x2F00, 0x2FD5) \
+		BIG_(0x2FF0, 0x2FFB) \
+		BIG_(0x3000, 0x3004) \
+		BIG_(0x3008, 0x3020) \
+		PAIR(0x3030, 0x3030) \
+		PAIR(0x3036, 0x3037) \
+		PAIR(0x303D, 0x303D) \
+		PAIR(0x303E, 0x303F) \
+		PAIR(0x309B, 0x309C) \
+		PAIR(0x30A0, 0x30A0) \
+		PAIR(0x30FB, 0x30FB) \
+		BIG_(0x31C0, 0x31E3) \
+		PAIR(0x321D, 0x321E) \
+		BIG_(0x3250, 0x325F) \
+		PAIR(0x327C, 0x327E) \
+		BIG_(0x32B1, 0x32BF) \
+		PAIR(0x32CC, 0x32CF) \
+		PAIR(0x3377, 0x337A) \
+		PAIR(0x33DE, 0x33DF) \
+		PAIR(0x33FF, 0x33FF) \
+		BIG_(0x4DC0, 0x4DFF) \
+		BIG_(0xA490, 0xA4C6) \
+		BIG_(0xA60D, 0xA60F) \
+		BIG_(0xA673, 0xA673) \
+		BIG_(0xA67E, 0xA67F) \
+		BIG_(0xA700, 0xA721) \
+		BIG_(0xA788, 0xA788) \
+		BIG_(0xA828, 0xA82B) \
+		BIG_(0xA838, 0xA839) \
+		BIG_(0xA874, 0xA877) \
+		BIG_(0xFB29, 0xFB29) \
+		BIG_(0xFD3E, 0xFD3F) \
+		BIG_(0xFDFD, 0xFDFD) \
+		BIG_(0xFE10, 0xFE19) \
+		BIG_(0xFE30, 0xFE52) \
+		BIG_(0xFE54, 0xFE66) \
+		BIG_(0xFE68, 0xFE6B) \
+		BIG_(0xFF01, 0xFF20) \
+		BIG_(0xFF3B, 0xFF40) \
+		BIG_(0xFF5B, 0xFF65) \
+		BIG_(0xFFE0, 0xFFE6) \
+		BIG_(0xFFE8, 0xFFEE) \
+		BIG_(0xFFF9, 0xFFFD)
+		/*
+		{0x10101, 0x10101},
+		{0x10140, 0x1019B},
+		{0x1091F, 0x1091F},
+		{0x10B39, 0x10B3F},
+		{0x10E60, 0x10E7E},
+		{0x1D200, 0x1D241},
+		{0x1D245, 0x1D245},
+		{0x1D300, 0x1D356},
+		{0x1D6DB, 0x1D6DB},
+		{0x1D715, 0x1D715},
+		{0x1D74F, 0x1D74F},
+		{0x1D789, 0x1D789},
+		{0x1D7C3, 0x1D7C3},
+		{0x1D7CE, 0x1D7FF},
+		{0x1F000, 0x1F02B},
+		{0x1F030, 0x1F093},
+		{0x1F100, 0x1F10A}
+		*/
+		ARRAY
+#  undef BIG_
+#  undef PAIR
+	};
+#  define BIG_(a,b)
+#  define PAIR(a,b) (a << 2) | (b-a),
+	static const uint16_t neutral_p[] = { ARRAY };
+#  undef BIG_
+#  undef PAIR
+#  define BIG_(a,b) char big_##a[b < 0x4000 && b-a <= 3 ? -1 : 1];
+#  define PAIR(a,b) char pair##a[b >= 0x4000 || b-a > 3 ? -1 : 1];
+	struct CHECK { ARRAY };
+#  undef BIG_
+#  undef PAIR
+#  undef ARRAY
+
+	if (in_interval_table(wc, neutral_b, ARRAY_SIZE(neutral_b) - 1))
+		return 1;
+	if (in_uint16_table(wc, neutral_p, ARRAY_SIZE(neutral_p) - 1))
+		return 1;
+	return 0;
+}
+#  endif
+
 # endif /* UNICODE_BIDI_SUPPORT */
 
 #endif /* Homegrown Unicode support */
diff --git a/libbb/unicode_wcwidth.c b/libbb/unicode_wcwidth.c
index 7eccc394c..0bb622705 100644
--- a/libbb/unicode_wcwidth.c
+++ b/libbb/unicode_wcwidth.c
@@ -538,6 +538,6 @@ static int wcwidth(unsigned ucs)
 		|| ((ucs >> 17) == (2 >> 1)) /* 20000..3ffff: Supplementary and Tertiary Ideographic Planes */
 #  endif
 		);
-# endif
-#endif
+# endif /* >= 0x1100 */
+#endif /* >= 0x300 */
 }