Bug 1428771 - Make UCS2_CHAR_IS_BIDI check for lead surrogates corresponding to U+1E800...U+1EFFF and rename to UTF16_CODE_UNIT_IS_BIDI. r=jfkthame

MozReview-Commit-ID: 9ZKF6SaN79n

--HG--
extra : rebase_source : 406f2647e65e2bdb97190b12564bafa6df39e6a8

browser/base/content/browser.js

@@ -7484,11 +7484,11 @@ var gIdentityHandler = {
 
         // If the organization name starts with an RTL character, then
         // swap the positions of the organization and country code labels.
-        // The Unicode ranges reflect the definition of the UCS2_CHAR_IS_BIDI
+        // The Unicode ranges reflect the definition of the UTF16_CODE_UNIT_IS_BIDI
         // macro in intl/unicharutil/util/nsBidiUtils.h. When bug 218823 gets
         // fixed, this test should be replaced by one adhering to the
         // Unicode Bidirectional Algorithm proper (at the paragraph level).
-        icon_labels_dir = /^[\u0590-\u08ff\ufb1d-\ufdff\ufe70-\ufefc]/.test(icon_label) ?
+        icon_labels_dir = /^[\u0590-\u08ff\ufb1d-\ufdff\ufe70-\ufefc\ud802\ud803\ud83a\ud83b]/.test(icon_label) ?
                           "rtl" : "ltr";
       }
     } else if (this._pageExtensionPolicy) {

intl/unicharutil/util/nsBidiUtils.h

@@ -263,9 +263,17 @@ typedef enum nsCharType nsCharType;
                                      ((0xfe70 <= (c)) && ((c) <= 0xfefc)))
 #define IS_IN_SMP_RTL_BLOCK(c) (((0x10800 <= (c)) && ((c) <= 0x10fff)) || \
                                 ((0x1e800 <= (c)) && ((c) <= 0x1eFFF)))
-#define UCS2_CHAR_IS_BIDI(c) ((IS_IN_BMP_RTL_BLOCK(c)) || \
-                              (IS_RTL_PRESENTATION_FORM(c)) || \
-                              (c) == 0xD802 || (c) == 0xD803)
+// Due to the supplementary-plane RTL blocks being identifiable from the
+// high surrogate without examining the low surrogate, it is correct to
+// use this by-code-unit check on potentially astral text without doing
+// the math to decode surrogate pairs into code points. However, unpaired
+// high surrogates that are RTL high surrogates then count as RTL even
+// though, if replaced by the REPLACEMENT CHARACTER, it would not be
+// RTL.
+#define UTF16_CODE_UNIT_IS_BIDI(c) ((IS_IN_BMP_RTL_BLOCK(c)) || \
+                                    (IS_RTL_PRESENTATION_FORM(c)) || \
+                                    (c) == 0xD802 || (c) == 0xD803 || \
+                                    (c) == 0xD83A || (c) == 0xD83B)
 #define UTF32_CHAR_IS_BIDI(c)  ((IS_IN_BMP_RTL_BLOCK(c)) || \
                                (IS_RTL_PRESENTATION_FORM(c)) || \
                                (IS_IN_SMP_RTL_BLOCK(c)))

layout/xul/nsTextBoxFrame.cpp

@@ -701,7 +701,7 @@ nsTextBoxFrame::CalculateTitleForWidth(gfxContext&          aRenderingContext,
                     break;
                 }
 
-                if (UCS2_CHAR_IS_BIDI(*pos)) {
+                if (UTF16_CODE_UNIT_IS_BIDI(*pos)) {
                     AddStateBits(NS_FRAME_IS_BIDI);
                 }
                 pos = nextPos;
@@ -738,7 +738,7 @@ nsTextBoxFrame::CalculateTitleForWidth(gfxContext&          aRenderingContext,
                     break;
                 }
 
-                if (UCS2_CHAR_IS_BIDI(*pos)) {
+                if (UTF16_CODE_UNIT_IS_BIDI(*pos)) {
                     AddStateBits(NS_FRAME_IS_BIDI);
                 }
                 prevPos = pos;
@@ -790,7 +790,7 @@ nsTextBoxFrame::CalculateTitleForWidth(gfxContext&          aRenderingContext,
                     break;
                 }
 
-                if (UCS2_CHAR_IS_BIDI(*leftPos)) {
+                if (UTF16_CODE_UNIT_IS_BIDI(*leftPos)) {
                     AddStateBits(NS_FRAME_IS_BIDI);
                 }
 
@@ -812,7 +812,7 @@ nsTextBoxFrame::CalculateTitleForWidth(gfxContext&          aRenderingContext,
                     break;
                 }
 
-                if (UCS2_CHAR_IS_BIDI(*pos)) {
+                if (UTF16_CODE_UNIT_IS_BIDI(*pos)) {
                     AddStateBits(NS_FRAME_IS_BIDI);
                 }
 

widget/cocoa/TextInputHandler.mm

@@ -744,7 +744,7 @@ TISInputSourceWrapper::IsForRTLLanguage()
     bool ret = TranslateToString(kVK_ANSI_A, 0, eKbdType_ANSI, str);
     NS_ENSURE_TRUE(ret, ret);
     char16_t ch = str.IsEmpty() ? char16_t(0) : str.CharAt(0);
-    mIsRTL = UCS2_CHAR_IS_BIDI(ch);
+    mIsRTL = UTF16_CODE_UNIT_IS_BIDI(ch);
   }
   return mIsRTL != 0;
 }