Stop combining text runs which are connected by 'COMMON' blocks.

ui/gfx/render_text_unittest.cc

 // Copyright (c) 2012 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 "ui/gfx/render_text.h"
 
 #include <algorithm>
 
 #include "base/format_macros.h"
 #include "base/i18n/break_iterator.h"
 #include "base/memory/scoped_ptr.h"
 #include "base/strings/string_split.h"
 #include "base/strings/string_util.h"
 #include "base/strings/stringprintf.h"
 #include "base/strings/utf_string_conversions.h"
 #include "testing/gtest/include/gtest/gtest.h"
 #include "third_party/skia/include/core/SkSurface.h"
 #include "ui/gfx/break_list.h"
 #include "ui/gfx/canvas.h"
 #include "ui/gfx/color_utils.h"
 #include "ui/gfx/font.h"
 #include "ui/gfx/range/range.h"
 #include "ui/gfx/range/range_f.h"
 #include "ui/gfx/render_text_harfbuzz.h"
+#include "ui/gfx/text_utils.h"
 
 #if defined(OS_WIN)
 #include "base/win/windows_version.h"
 #include "ui/gfx/platform_font_win.h"
 #endif
 
 #if defined(OS_MACOSX)
 #include <ApplicationServices/ApplicationServices.h>
 
 #include "ui/gfx/render_text_mac.h"
@@ skipping 48 matching lines @@
     render_text->MoveCursor(CHARACTER_BREAK, direction, false);
   }
   // Check that cursoring is clamped at the line edge.
   EXPECT_EQ(expected.back(), render_text->selection_model());
   // Check that it is the line edge.
   render_text->MoveCursor(LINE_BREAK, direction, false);
   EXPECT_EQ(expected.back(), render_text->selection_model());
 }
 #endif  // !defined(OS_MACOSX)
 
-// Test utility for Multiline_Newline test case. Empty |expected_range| means
-// the blank line which has no segments. Otherwise |segments| should contain
-// exactly one line segment whose range equals to |expected_range|.
-void VerifyLineSegments(const Range& expected_range,
-                        const std::vector<internal::LineSegment>& segments) {
-  EXPECT_EQ(expected_range.is_empty() ? 0ul : 1ul, segments.size());
-  if (!expected_range.is_empty())
-    EXPECT_EQ(expected_range, segments[0].char_range);
-}
-
 // The class which records the drawing operations so that the test case can
 // verify where exactly the glyphs are drawn.
 class TestSkiaTextRenderer : public internal::SkiaTextRenderer {
  public:
   struct TextLog {
     TextLog() : glyph_count(0u) {}
     PointF origin;
     size_t glyph_count;
   };
 
@@ skipping 2008 matching lines @@
     const wchar_t* const text;
     const Range first_line_char_range;
     const Range second_line_char_range;
     bool is_ltr;
   } kTestStrings[] = {
     { L"abc defg hijkl", Range(0, 9), Range(9, 14), true },
     { L"qwertyzxcvbn", Range(0, 10), Range(10, 12), true },
     { L"\x0627\x0644\x0644\x063A\x0629 "
       L"\x0627\x0644\x0639\x0631\x0628\x064A\x0629",
       Range(0, 6), Range(6, 13), false },
-    { L"\x062A\x0641\x0627\x062D\x05EA\x05E4\x05D5\x05D6\x05D9"
-      L"\x05DA\x05DB\x05DD", Range(0, 4), Range(4, 12), false }
+    { L"\x062A\x0641\x0627\x062D \x05EA\x05E4\x05D5\x05D6\x05D9"
+      L"\x05DA\x05DB\x05DD", Range(0, 5), Range(5, 13), false }
   };
 
   RenderTextHarfBuzz render_text;
   // Specify the fixed width for characters to suppress the possible variations
   // of linebreak results.
   render_text.set_glyph_width_for_test(5);
   render_text.SetDisplayRect(Rect(50, 1000));
   render_text.SetMultiline(true);
   render_text.SetWordWrapBehavior(WRAP_LONG_WORDS);
   render_text.SetHorizontalAlignment(ALIGN_TO_HEAD);
@@ skipping 79 matching lines @@
   for (size_t i = 0; i < arraysize(kTestStrings); ++i) {
     SCOPED_TRACE(base::StringPrintf("kTestStrings[%" PRIuS "]", i));
     render_text.SetText(WideToUTF16(kTestStrings[i].text));
     render_text.Draw(&canvas);
     EXPECT_EQ(kTestStrings[i].lines_count, render_text.lines_.size());
     if (kTestStrings[i].lines_count != render_text.lines_.size())
       continue;
 
     for (size_t j = 0; j < kTestStrings[i].lines_count; ++j) {
       SCOPED_TRACE(base::StringPrintf("Line %" PRIuS "", j));
-      VerifyLineSegments(kTestStrings[i].line_char_ranges[j],
-                         render_text.lines_[j].segments);
+      // There might be multiple segments in one line. Merge all the segments
+      // ranges in the same line.
+      const size_t segment_size = render_text.lines()[j].segments.size();
+      Range line_range;
+      if (segment_size > 0)
+        line_range = Range(
+            render_text.lines()[j].segments[0].char_range.start(),
+            render_text.lines()[j].segments[segment_size - 1].char_range.end());
+      EXPECT_EQ(kTestStrings[i].line_char_ranges[j], line_range);
     }
   }
 }
 
 // Make sure that multiline mode ignores elide behavior.
 TEST_F(RenderTextTest, Multiline_IgnoreElide) {
   const wchar_t kTestString[] =
       L"very very very long string xxxxxxxxxxxxxxxxxxxxxxxxxx";
   const wchar_t kEllipsis[] = L"\x2026";
 
@@ skipping 114 matching lines @@
     for (size_t j = 0; j < render_text.lines().size(); ++j) {
       SCOPED_TRACE(base::StringPrintf("%" PRIuS "-th line", j));
       EXPECT_EQ(kTestScenarios[i].char_ranges[j],
                 render_text.lines()[j].segments[0].char_range);
       EXPECT_EQ(kTestScenarios[i].char_ranges[j].length() * kGlyphSize,
                 render_text.lines()[j].size.width());
     }
   }
 }
 
+TEST_F(RenderTextTest, Multiline_LineBreakerBehavior) {
+  const int kGlyphSize = 5;
+  const struct {
+    const wchar_t* const text;
+    const WordWrapBehavior behavior;
+    const Range char_ranges[3];
+  } kTestScenarios[] = {
+      { L"a single run", IGNORE_LONG_WORDS,
+        {Range(0, 2), Range(2, 9), Range(9, 12) } },
+      // 3 words: "That's ", ""good". ", "aaa" and 7 runs: "That", "'", "s ",
+      // """, "good", "". ", "aaa". They all mixed together.
+      { L"That's \"good\". aaa", IGNORE_LONG_WORDS,
+        {Range(0, 7), Range(7, 15), Range(15, 18) } },
+      // Test "\"" should be put into a new line correctly.
+      { L"a \"good\" one.", IGNORE_LONG_WORDS,
+        {Range(0, 2), Range(2, 9), Range(9, 13) } },
+      // Test for full-width space.
+      { L"That's\x3000good.\x3000yyy", IGNORE_LONG_WORDS,
+        {Range(0, 7), Range(7, 13), Range(13, 16) } },
+      { L"a single run", TRUNCATE_LONG_WORDS,
+        {Range(0, 2), Range(2, 6), Range(9, 12) } },
+      { L"That's \"good\". aaa", TRUNCATE_LONG_WORDS,
+        {Range(0, 4), Range(7, 11), Range(15, 18) } },
+      { L"That's good. aaa", TRUNCATE_LONG_WORDS,
+        {Range(0, 4), Range(7, 11), Range(13, 16) } },
+      { L"a \"good\" one.", TRUNCATE_LONG_WORDS,
+        {Range(0, 2), Range(2, 6), Range(9, 13) } },
+      { L"asingleword", WRAP_LONG_WORDS,
+        {Range(0, 4), Range(4, 8), Range(8, 11) } },
+      { L"That's good", WRAP_LONG_WORDS,
+        {Range(0, 4), Range(4, 7), Range(7, 11) } },
+      { L"That's \"g\".", WRAP_LONG_WORDS,
+        {Range(0, 4), Range(4, 7), Range(7, 11) } },
+  };
+
+  RenderTextHarfBuzz render_text;
+  render_text.SetMultiline(true);
+  render_text.set_glyph_width_for_test(kGlyphSize);
+  render_text.SetDisplayRect(Rect(0, 0, kGlyphSize * 4, 0));
+
+  Canvas canvas;
+
+  for (size_t i = 0; i < arraysize(kTestScenarios); ++i) {
+    SCOPED_TRACE(base::StringPrintf("kTestStrings[%" PRIuS "]", i));
+    render_text.SetText(WideToUTF16(kTestScenarios[i].text));
+    render_text.SetWordWrapBehavior(kTestScenarios[i].behavior);
+    render_text.Draw(&canvas);
+
+    ASSERT_EQ(3u, render_text.lines().size());
+    for (size_t j = 0; j < render_text.lines().size(); ++j) {
+      SCOPED_TRACE(base::StringPrintf("%" PRIuS "-th line", j));
+      // Merge all the segments ranges in the same line.
+      size_t segment_size = render_text.lines()[j].segments.size();
+      Range line_range;
+      if (segment_size > 0)
+        line_range = Range(
+            render_text.lines()[j].segments[0].char_range.start(),
+            render_text.lines()[j].segments[segment_size - 1].char_range.end());
+      EXPECT_EQ(kTestScenarios[i].char_ranges[j], line_range);
+      EXPECT_EQ(kTestScenarios[i].char_ranges[j].length() * kGlyphSize,
+                render_text.lines()[j].size.width());
+    }
+  }
+}
+
+// Test that Surrogate pairs or combining character sequences do not get
+// separated by line breaking.
+TEST_F(RenderTextTest, Multiline_SurrogatePairsOrCombiningChars) {
+  RenderTextHarfBuzz render_text;
+  render_text.SetMultiline(true);
+  render_text.SetWordWrapBehavior(WRAP_LONG_WORDS);
+
+  // Below is 'MUSICAL SYMBOL G CLEF' (U+1D11E), which is represented in UTF-16
+  // as two code units forming a surrogate pair: 0xD834 0xDD1E.
+  const base::char16 kSurrogate[] = {0xD834, 0xDD1E, 0};
+  // Below is a Devanagari two-character combining sequence U+0921 U+093F. The
+  // sequence forms a single display character and should not be separated.
+  const base::char16 kCombiningChars[] = {0x921, 0x93F, 0};
+  const float kSurrogateWidth =
+      GetStringWidthF(kSurrogate, render_text.font_list());
+  const float kCombiningCharsWidth =
+      GetStringWidthF(kCombiningChars, render_text.font_list());
+
+  // Construct two words.
+  const base::string16 text(kSurrogate + UTF8ToUTF16(" ") + kCombiningChars);
+  // Carefully set display rectangle width so that it will split the surrogate
+  // pairs and combining characters.
+  const int width = std::min(kSurrogateWidth, kCombiningCharsWidth) / 2;
+  const Range char_ranges[3] = {Range(0, 2), Range(2, 3), Range(3, 5)};
+
+  Canvas canvas;
+  render_text.SetText(text);
+  render_text.SetDisplayRect(Rect(0, 0, width, 0));
+  render_text.Draw(&canvas);
+
+  ASSERT_EQ(3u, render_text.lines().size());
+  for (size_t j = 0; j < render_text.lines().size(); ++j) {
+    SCOPED_TRACE(base::StringPrintf("%" PRIuS "-th line", j));
+    // There is only one segment in each line.
+    EXPECT_EQ(char_ranges[j], render_text.lines()[j].segments[0].char_range);
+  }
+}
+
+// Test that Zero width characters have the correct line breaking behavior.
+TEST_F(RenderTextTest, Multiline_ZeroWidthChars) {
+  RenderTextHarfBuzz render_text;
+  render_text.SetMultiline(true);
+  render_text.SetWordWrapBehavior(WRAP_LONG_WORDS);
+  render_text.SetFontList(FontList("Arial, 13px"));

[msw, 2015/06/09 01:15:50]

Avoid relying on a particular fonts, it might break on Win XP, etc.

[xdai1, 2015/06/09 17:59:21]

Done.

+
+  const base::char16 kZeroWidthSpace = {0x200B};
+  const base::string16 text(UTF8ToUTF16("test") + kZeroWidthSpace +
+                            kZeroWidthSpace + UTF8ToUTF16("test."));
+  const int kTestWidth =
+      GetStringWidth(UTF8ToUTF16("test"), render_text.font_list());
+  const Range char_ranges[3] = {Range(0, 6), Range(6, 10), Range(10, 11)};

[msw, 2015/06/09 01:15:50]

My concern was actually for zero width spaces at the start of a line, but I
guess that's pretty obscure if it's even possible ("\n" + kZeroWidthSpace?)...

[xdai1, 2015/06/09 17:59:21]

I'm not quite understand your concern... Why would "\n" + kZeroWidthSpace cause
any problem? I modified the test RenderTextTest.Multiline_ZeroWidthChars to
include the case that zero width space at the start of a line. Is that what you
mean?

[msw, 2015/06/09 22:51:53]

I think my concern was that the line breaker might put a zero-width space on its
own line... but I feel like this concern isn't worth investigating further at
this point.

+
+  Canvas canvas;
+  render_text.SetText(text);
+  render_text.SetDisplayRect(Rect(0, 0, kTestWidth, 0));
+  render_text.Draw(&canvas);
+
+  ASSERT_EQ(3u, render_text.lines().size());
+  for (size_t j = 0; j < render_text.lines().size(); ++j) {
+    SCOPED_TRACE(base::StringPrintf("%" PRIuS "-th line", j));
+    // Merge all the segments ranges in the same line.
+    size_t segment_size = render_text.lines()[j].segments.size();
+    Range line_range;
+    if (segment_size > 0)

[msw, 2015/06/09 01:15:50]

nit: instead ASSERT_GT(segment_size, 0), and "Range line_range(render_text..."

[xdai1, 2015/06/09 17:59:21]

Done.

+      line_range = Range(
+          render_text.lines()[j].segments[0].char_range.start(),
+          render_text.lines()[j].segments[segment_size - 1].char_range.end());
+    // There is only one segment in each line.
+    EXPECT_EQ(char_ranges[j], line_range);
+  }
+}
+
 TEST_F(RenderTextTest, NewlineWithoutMultilineFlag) {
   const wchar_t* kTestStrings[] = {
     L"abc\ndef", L"a \n b ", L"ab\n", L"a\n\nb", L"\nab", L"\n",
   };
 
   RenderTextHarfBuzz render_text;
   render_text.SetDisplayRect(Rect(200, 1000));
   Canvas canvas;
 
   for (size_t i = 0; i < arraysize(kTestStrings); ++i) {
@@ skipping 225 matching lines @@
   render_text.EnsureLayout();
   internal::TextRunList* run_list = render_text.GetRunList();
   ASSERT_EQ(3U, run_list->size());
   EXPECT_EQ(Range(0, 1), run_list->runs()[0]->range);
   EXPECT_EQ(Range(1, 2), run_list->runs()[1]->range);
   EXPECT_EQ(Range(2, 3), run_list->runs()[2]->range);
 
   render_text.SetText(WideToUTF16(L"x \x25B6 y"));
   render_text.EnsureLayout();
   run_list = render_text.GetRunList();
-  ASSERT_EQ(3U, run_list->size());
+  ASSERT_EQ(4U, run_list->size());
   EXPECT_EQ(Range(0, 2), run_list->runs()[0]->range);
   EXPECT_EQ(Range(2, 3), run_list->runs()[1]->range);
-  EXPECT_EQ(Range(3, 5), run_list->runs()[2]->range);
+  EXPECT_EQ(Range(3, 4), run_list->runs()[2]->range);
+  EXPECT_EQ(Range(4, 5), run_list->runs()[3]->range);
 }
 
 TEST_F(RenderTextTest, HarfBuzz_BreakRunsByEmoji) {
   RenderTextHarfBuzz render_text;
 
   // \xF0\x9F\x98\x81 (U+1F601) is smile icon emoji. \xE2\x9C\xA8 (U+2728) is
   // a sparkle icon. Both can be drawn with color emoji fonts, so runs should be
   // separated. See crbug.com/448909
   render_text.SetText(UTF8ToUTF16("x\xF0\x9F\x98\x81y\xE2\x9C\xA8"));
   render_text.EnsureLayout();
@@ skipping 314 matching lines @@
   string_size.set_width(string_size.width() / 2);
   render_text.SetDisplayRect(gfx::Rect(string_size));
   render_text.EnsureLayout();
   CFIndex glyph_count = CTLineGetGlyphCount(render_text.line_);
   EXPECT_GT(text.size(), static_cast<size_t>(glyph_count));
   EXPECT_NE(0, glyph_count);
 }
 #endif
 
 }  // namespace gfx