Changeset 121756 in webkit

Timestamp:

Jul 3, 2012 3:29:08 AM (12 years ago)

Author:

pdr@google.com

Message:

Fix text positioning with non-bmp characters.
​https://bugs.webkit.org/show_bug.cgi?id=87681

Reviewed by Nikolas Zimmermann.

Source/WebCore:

Previously when constructing metrics for tspans with non-bmp characters,
each non-bmp character treated as a skipped character in the same way that
spaces are ignored.
This made sense because the initial SVGCharacterDataMap for <text> is
indexed by character index (not string length) so the high portion of a
non-bmp character was treated as a skipped space. Unfortunately, this
led to a bug because skipped spaces lead to an offset in the positioning
values list but non-bmp characters do not.

This change switches the code to use a new offset for non-bmp characters,
surrogatePairCharacters, which does not affect the positioning values list.

Tests: svg/text/non-bmp-tspans-expected.svg

svg/text/non-bmp-tspans.svg

• rendering/svg/SVGTextMetricsBuilder.cpp:

(WebCore::SVGTextMetricsBuilder::measureTextRenderer):

LayoutTests:

• svg/text/non-bmp-tspans-expected.svg: Added.
• svg/text/non-bmp-tspans.svg: Added.

Location:

trunk

Files:

• LayoutTests/ChangeLog (modified) (1 diff)
• LayoutTests/svg/text/non-bmp-tspans-expected.svg (added)
• LayoutTests/svg/text/non-bmp-tspans.svg (added)
• Source/WebCore/ChangeLog (modified) (1 diff)
• Source/WebCore/rendering/svg/SVGTextMetricsBuilder.cpp (modified) (3 diffs)

trunk/LayoutTests/ChangeLog

@@ +1 @@
+2012-07-03  Philip Rogers  <pdr@google.com>
+
+        Fix text positioning with non-bmp characters.
+        https://bugs.webkit.org/show_bug.cgi?id=87681
+
+        Reviewed by Nikolas Zimmermann.
+
+        * svg/text/non-bmp-tspans-expected.svg: Added.
+        * svg/text/non-bmp-tspans.svg: Added.
+
 2012-07-03  Gyuyoung Kim  <gyuyoung.kim@samsung.com>
 

trunk/Source/WebCore/ChangeLog

@@ +1 @@
+2012-07-03  Philip Rogers  <pdr@google.com>
+
+        Fix text positioning with non-bmp characters.
+        https://bugs.webkit.org/show_bug.cgi?id=87681
+
+        Reviewed by Nikolas Zimmermann.
+
+        Previously when constructing metrics for tspans with non-bmp characters,
+        each non-bmp character treated as a skipped character in the same way that
+        spaces are ignored.
+        This made sense because the initial SVGCharacterDataMap for <text> is
+        indexed by character index (not string length) so the high portion of a
+        non-bmp character was treated as a skipped space. Unfortunately, this
+        led to a bug because skipped spaces lead to an offset in the positioning
+        values list but non-bmp characters do not.
+
+        This change switches the code to use a new offset for non-bmp characters,
+        surrogatePairCharacters, which does not affect the positioning values list.
+
+        Tests: svg/text/non-bmp-tspans-expected.svg
+               svg/text/non-bmp-tspans.svg
+
+        * rendering/svg/SVGTextMetricsBuilder.cpp:
+        (WebCore::SVGTextMetricsBuilder::measureTextRenderer):
+
 2012-07-03  Gyuyoung Kim  <gyuyoung.kim@samsung.com>
 

trunk/Source/WebCore/rendering/svg/SVGTextMetricsBuilder.cpp

@@ -156 +156 @@
     initializeMeasurementWithTextRenderer(text);
     bool preserveWhiteSpace = text->style()->whiteSpace() == PRE;
+    int surrogatePairCharacters = 0;
 
     while (advance()) {
@@ -169 +170 @@
         if (data->processRenderer) {
             if (data->allCharactersMap) {
-                const SVGCharacterDataMap::const_iterator it = data->allCharactersMap->find(data->valueListPosition + m_textPosition - data->skippedCharacters + 1);
+                const SVGCharacterDataMap::const_iterator it = data->allCharactersMap->find(data->valueListPosition + m_textPosition - data->skippedCharacters - surrogatePairCharacters + 1);
                 if (it != data->allCharactersMap->end())
                     attributes->characterDataMap().set(m_textPosition + 1, it->second);
@@ -177 +178 @@
 
         if (data->allCharactersMap && currentCharacterStartsSurrogatePair())
-            data->skippedCharacters += m_currentMetrics.length() - 1;
+            surrogatePairCharacters++;
 
         data->lastCharacter = currentCharacter;