Add a SymbolsIterator to aid with Symbols Segmentation

third_party/WebKit/Source/platform/fonts/SymbolsIterator.cpp

Index: third_party/WebKit/Source/platform/fonts/SymbolsIterator.cpp
diff --git a/third_party/WebKit/Source/platform/fonts/SymbolsIterator.cpp b/third_party/WebKit/Source/platform/fonts/SymbolsIterator.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..475841095274c92d61f8f71c49f706f14ce26bd1
--- /dev/null
+++ b/third_party/WebKit/Source/platform/fonts/SymbolsIterator.cpp
@@ -0,0 +1,145 @@
+// Copyright 2015 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "SymbolsIterator.h"
+
+#include <unicode/uchar.h>
+#include <unicode/uniset.h>
+
+namespace blink {
+
+using namespace WTF::Unicode;
+
+SymbolsIterator::SymbolsIterator(const UChar* buffer, unsigned bufferSize)
+    : m_utf16Iterator(adoptPtr(new UTF16TextIterator(buffer, bufferSize)))
+    , m_bufferSize(bufferSize)
+    , m_nextChar(0)
+    , m_atEnd(bufferSize == 0)
+    , m_currentFontFallbackPriority(FontFallbackPriority::Invalid)
+{
+}
+
+FontFallbackPriority SymbolsIterator::fontFallbackPriorityForCharacter(UChar32 codepoint)
+{
+
+    // Those should only be Emoji presentation as combinations of two.
+    if (Character::isEmojiKeycapBase(codepoint) || Character::isRegionalIndicator(codepoint))
+        return FontFallbackPriority::Text;
+
+    if (codepoint == 0x20E3 || codepoint == 0x20E0)

[eae, 2016/01/22 00:38:32]

Could you please add these codepoints to CharacterNames.h and use the constants
instead?

+        return FontFallbackPriority::EmojiEmoji;
+
+    // There are overlaps between isEmojiTextpresentation and isEmojiEmojiPresentation.
+    // We want to prioritize emoji presentation if the character has such a property.
+    if (Character::isEmojiEmojiPresentation(codepoint)
+        || Character::isEmojiModifierBase(codepoint)
+        || Character::isModifier(codepoint))
+        return FontFallbackPriority::EmojiEmoji;
+
+    if (Character::isEmojiTextPresentation(codepoint))
+        return FontFallbackPriority::EmojiText;
+
+    UBlockCode block = ublock_getCode(codepoint);
+
+    switch (block) {
+    case UBLOCK_PLAYING_CARDS:
+    case UBLOCK_MISCELLANEOUS_SYMBOLS:
+    case UBLOCK_MISCELLANEOUS_SYMBOLS_AND_ARROWS:
+    case UBLOCK_MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS:
+    case UBLOCK_TRANSPORT_AND_MAP_SYMBOLS:
+    case UBLOCK_ALCHEMICAL_SYMBOLS:
+    case UBLOCK_RUNIC:
+    case UBLOCK_DINGBATS:
+    case UBLOCK_GOTHIC:
+        return FontFallbackPriority::Symbols;
+    case UBLOCK_ARROWS:
+    case UBLOCK_MATHEMATICAL_OPERATORS:
+    case UBLOCK_MISCELLANEOUS_TECHNICAL:
+    case UBLOCK_GEOMETRIC_SHAPES:
+    case UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A:
+    case UBLOCK_SUPPLEMENTAL_ARROWS_A:
+    case UBLOCK_SUPPLEMENTAL_ARROWS_B:
+    case UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B:
+    case UBLOCK_SUPPLEMENTAL_MATHEMATICAL_OPERATORS:
+    case UBLOCK_MATHEMATICAL_ALPHANUMERIC_SYMBOLS:
+    case UBLOCK_ARABIC_MATHEMATICAL_ALPHABETIC_SYMBOLS:
+    case UBLOCK_GEOMETRIC_SHAPES_EXTENDED:
+        return FontFallbackPriority::Math;
+    default:
+        return FontFallbackPriority::Text;
+    }
+}
+
+bool SymbolsIterator::consume(unsigned *symbolsLimit, FontFallbackPriority* fontFallbackPriority)
+{
+    if (m_atEnd)
+        return false;
+
+    while (m_utf16Iterator->consume(m_nextChar)) {
+        m_previousFontFallbackPriority = m_currentFontFallbackPriority;
+        unsigned iteratorOffset = m_utf16Iterator->offset();
+        m_utf16Iterator->advance();
+
+        // ZWJ just carries over the emoji or neutral text type, VS15 & VS16 we just carry over as well, since we
+        // are resolved those through lookahead - except at the beginning.
+        // Also, the text presentation Keep ZWJ joiner pairs in Emoji presentation. See below.
+        // Example U+1F441 U+200D U+1F5E8, eye + ZWJ + left speech bubble
+        if ((!(m_nextChar == zeroWidthJoinerCharacter && m_previousFontFallbackPriority == FontFallbackPriority::EmojiEmoji)
+            && m_nextChar != variationSelector15Character
+            && m_nextChar != variationSelector16Character
+            && !Character::isRegionalIndicator(m_nextChar)
+            && !(m_nextChar == 0x1F5E8 && m_previousFontFallbackPriority == FontFallbackPriority::EmojiEmoji))
+            || m_currentFontFallbackPriority == FontFallbackPriority::Invalid) {
+            m_currentFontFallbackPriority = fontFallbackPriorityForCharacter(m_nextChar);
+        }
+
+        UChar32 peekChar = 0;
+        if (m_utf16Iterator->consume(peekChar)
+            && peekChar != 0) {
+
+            // Variation Selectors
+            if (m_currentFontFallbackPriority == FontFallbackPriority::EmojiEmoji && peekChar == variationSelector15Character) {
+                m_currentFontFallbackPriority = FontFallbackPriority::EmojiText;
+            }
+
+            if (m_currentFontFallbackPriority == FontFallbackPriority::EmojiText && peekChar == variationSelector16Character) {
+                m_currentFontFallbackPriority = FontFallbackPriority::EmojiEmoji;
+            }
+
+            // Combining characters Keycap...
+            if (Character::isEmojiKeycapBase(m_nextChar) && peekChar == 0x20E3) {
+                m_currentFontFallbackPriority = FontFallbackPriority::EmojiEmoji;
+            };
+
+            // ...and Combining Enclosing Circle Backslash.
+            if (m_currentFontFallbackPriority == FontFallbackPriority::EmojiText && peekChar == 0x20E0) {
+                m_currentFontFallbackPriority = FontFallbackPriority::EmojiEmoji;
+            }
+
+            // Regional indicators
+            if (Character::isRegionalIndicator(m_nextChar) && Character::isRegionalIndicator(peekChar)) {
+                m_currentFontFallbackPriority = FontFallbackPriority::EmojiEmoji;
+            }
+
+            // Upgrade text presentation emoji to emoji presentation when followed by ZWJ,
+            // Example U+1F441 U+200D U+1F5E8, eye + ZWJ + left speech bubble
+            if (m_nextChar == 0x1F441 && peekChar == zeroWidthJoinerCharacter) {
+                m_currentFontFallbackPriority = FontFallbackPriority::EmojiEmoji;
+            }
+        }
+
+        if (m_previousFontFallbackPriority != m_currentFontFallbackPriority
+            && (m_previousFontFallbackPriority != FontFallbackPriority::Invalid)) {
+            *symbolsLimit = iteratorOffset;
+            *fontFallbackPriority = m_previousFontFallbackPriority;
+            return true;
+        }
+    }
+    *symbolsLimit = m_bufferSize;
+    *fontFallbackPriority = m_currentFontFallbackPriority;
+    m_atEnd = true;
+    return true;
+}
+
+}