bug 731536 - text-transform:capitalize should affect the first alphanumeric char in a word, not the first non-punctuation char. r=smontagu

content/base/src/nsLineBreaker.cpp

@@ -60,16 +60,26 @@ static void
 SetupCapitalization(const PRUnichar* aWord, PRUint32 aLength,
                     bool* aCapitalization)
 {
-  // Capitalize the first non-punctuation character after a space or start
+  // Capitalize the first alphanumeric character after a space or start
   // of the word.
   // The only space character a word can contain is NBSP.
   bool capitalizeNextChar = true;
   for (PRUint32 i = 0; i < aLength; ++i) {
-    if (capitalizeNextChar && !nsContentUtils::IsFirstLetterPunctuation(aWord[i])) {
-      aCapitalization[i] = true;
-      capitalizeNextChar = false;
+    PRUint32 ch = aWord[i];
+    if (capitalizeNextChar) {
+      if (NS_IS_HIGH_SURROGATE(ch) && i + 1 < aLength &&
+          NS_IS_LOW_SURROGATE(aWord[i + 1])) {
+        ch = SURROGATE_TO_UCS4(ch, aWord[i + 1]);
+      }
+      if (nsContentUtils::IsAlphanumeric(ch)) {
+        aCapitalization[i] = true;
+        capitalizeNextChar = false;
+      }
+      if (!IS_IN_BMP(ch)) {
+        ++i;
+      }
     }
-    if (aWord[i] == 0xA0 /*NBSP*/) {
+    if (ch == 0xA0 /*NBSP*/) {
       capitalizeNextChar = true;
     }
   }

layout/reftests/text-transform/all-title-ref.html

@@ -245,7 +245,7 @@
 Ʋx
 Ʌx
 Ʒx
-Ιx
+&#x0345;X<!-- combining Greek ypogegrammeni - category Mn, doesn't get titlecased -->
 &#x0370;x
 &#x0372;x
 &#x0376;x
@@ -732,32 +732,32 @@
 &#x216E;x
 &#x216F;x
 &#x2183;x
-&#x24B6;x
-&#x24B7;x
-&#x24B8;x
-&#x24B9;x
-&#x24BA;x
-&#x24BB;x
-&#x24BC;x
-&#x24BD;x
-&#x24BE;x
-&#x24BF;x
-&#x24C0;x
-&#x24C1;x
-&#x24C2;x
-&#x24C3;x
-&#x24C4;x
-&#x24C5;x
-&#x24C6;x
-&#x24C7;x
-&#x24C8;x
-&#x24C9;x
-&#x24CA;x
-&#x24CB;x
-&#x24CC;x
-&#x24CD;x
-&#x24CE;x
-&#x24CF;x
+&#x24D0;X<!-- circled Latin small letters - category So - not titlecased -->
+&#x24D1;X
+&#x24D2;X
+&#x24D3;X
+&#x24D4;X
+&#x24D5;X
+&#x24D6;X
+&#x24D7;X
+&#x24D8;X
+&#x24D9;X
+&#x24DA;X
+&#x24DB;X
+&#x24DC;X
+&#x24DD;X
+&#x24DE;X
+&#x24DF;X
+&#x24E0;X
+&#x24E1;X
+&#x24E2;X
+&#x24E3;X
+&#x24E4;X
+&#x24E5;X
+&#x24E6;X
+&#x24E7;X
+&#x24E8;X
+&#x24E9;X<!-- end of circled Latin small letters -->
 &#x2C00;x
 &#x2C01;x
 &#x2C02;x

layout/reftests/text-transform/all-title.html

@@ -248,7 +248,7 @@
 ʋx
 ʌx
 ʒx
-ͅx
+&#x0345;x<!-- combining Greek ypogegrammeni - category Mn, doesn't get titlecased -->
 &#x0371;x
 &#x0373;x
 &#x0377;x
@@ -735,7 +735,7 @@
 &#x217E;x
 &#x217F;x
 &#x2184;x
-&#x24D0;x
+&#x24D0;x<!-- circled Latin small letters - category So - not titlecased -->
 &#x24D1;x
 &#x24D2;x
 &#x24D3;x
@@ -760,7 +760,7 @@
 &#x24E6;x
 &#x24E7;x
 &#x24E8;x
-&#x24E9;x
+&#x24E9;x<!-- end of circled Latin small letters -->
 &#x2C30;x
 &#x2C31;x
 &#x2C32;x