mirrors/ladybird
0
0
Fork 0
mirror of https://github.com/LadybirdBrowser/ladybird.git synced 2025-08-28 13:18:19 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 9

LibWeb: Iterate over text chunks using a grapheme-aware segmenter

Browse source 
Our current text iterator is not aware of multi-code point graphemes.
Instead of simply incrementing an iterator one code point at a time, use
our Unicode grapheme segmenter to break text into fragments.
This commit is contained in:
Timothy Flynn 2024-09-20 17:14:14 -04:00 • committed by Andreas Kling
commit f0105b473b
Notes: github-actions[bot] 2024-09-21 06:58:43 +00:00
5 changed files with 84 additions and 32 deletions
Download patch file Download diff file Expand all files Collapse all files

69
Userland/Libraries/LibWeb/Layout/TextNode.cpp
View file

@ -391,13 +391,14 @@ void TextNode::compute_text_for_rendering()
m_text_for_rendering = MUST(builder.to_string());
}
TextNode::ChunkIterator::ChunkIterator(StringView text, bool wrap_lines, bool respect_linebreaks, Gfx::FontCascadeList const& font_cascade_list)
TextNode::ChunkIterator::ChunkIterator(String const& text, bool wrap_lines, bool respect_linebreaks, Gfx::FontCascadeList const& font_cascade_list)
: m_wrap_lines(wrap_lines)
, m_respect_linebreaks(respect_linebreaks)
, m_utf8_view(text)
, m_iterator(m_utf8_view.begin())
, m_font_cascade_list(font_cascade_list)
, m_segmenter(Unicode::Segmenter::create(Unicode::SegmenterGranularity::Grapheme))
{
m_segmenter->set_segmented_text(text);
}
static Gfx::GlyphRun::TextType text_type_for_code_point(u32 code_point)
@ -462,75 +463,85 @@ Optional<TextNode::Chunk> TextNode::ChunkIterator::peek(size_t count)
Optional<TextNode::Chunk> TextNode::ChunkIterator::next_without_peek()
{
if (m_iterator == m_utf8_view.end())
if (m_current_index >= m_utf8_view.byte_length())
return {};
auto start_of_chunk = m_iterator;
auto current_code_point = [this]() {
return *m_utf8_view.iterator_at_byte_offset_without_validation(m_current_index);
};
auto next_grapheme_boundary = [this]() {
return m_segmenter->next_boundary(m_current_index).value_or(m_utf8_view.byte_length());
};
Gfx::Font const& font = m_font_cascade_list.font_for_code_point(*m_iterator);
auto text_type = text_type_for_code_point(*m_iterator);
while (m_iterator != m_utf8_view.end()) {
if (&font != &m_font_cascade_list.font_for_code_point(*m_iterator)) {
if (auto result = try_commit_chunk(start_of_chunk, m_iterator, false, font, text_type); result.has_value())
auto code_point = current_code_point();
auto start_of_chunk = m_current_index;
Gfx::Font const& font = m_font_cascade_list.font_for_code_point(code_point);
auto text_type = text_type_for_code_point(code_point);
while (m_current_index < m_utf8_view.byte_length()) {
code_point = current_code_point();
if (&font != &m_font_cascade_list.font_for_code_point(code_point)) {
if (auto result = try_commit_chunk(start_of_chunk, m_current_index, false, font, text_type); result.has_value())
return result.release_value();
}
if (m_respect_linebreaks && *m_iterator == '\n') {
if (m_respect_linebreaks && code_point == '\n') {
// Newline encountered, and we're supposed to preserve them.
// If we have accumulated some code points in the current chunk, commit them now and continue with the newline next time.
if (auto result = try_commit_chunk(start_of_chunk, m_iterator, false, font, text_type); result.has_value())
if (auto result = try_commit_chunk(start_of_chunk, m_current_index, false, font, text_type); result.has_value())
return result.release_value();
// Otherwise, commit the newline!
++m_iterator;
auto result = try_commit_chunk(start_of_chunk, m_iterator, true, font, text_type);
m_current_index = next_grapheme_boundary();
auto result = try_commit_chunk(start_of_chunk, m_current_index, true, font, text_type);
VERIFY(result.has_value());
return result.release_value();
}
if (m_wrap_lines) {
if (text_type != text_type_for_code_point(*m_iterator)) {
if (auto result = try_commit_chunk(start_of_chunk, m_iterator, false, font, text_type); result.has_value())
if (text_type != text_type_for_code_point(code_point)) {
if (auto result = try_commit_chunk(start_of_chunk, m_current_index, false, font, text_type); result.has_value()) {
return result.release_value();
}
}
if (is_ascii_space(*m_iterator)) {
if (is_ascii_space(code_point)) {
// Whitespace encountered, and we're allowed to break on whitespace.
// If we have accumulated some code points in the current chunk, commit them now and continue with the whitespace next time.
if (auto result = try_commit_chunk(start_of_chunk, m_iterator, false, font, text_type); result.has_value())
if (auto result = try_commit_chunk(start_of_chunk, m_current_index, false, font, text_type); result.has_value()) {
return result.release_value();
}
// Otherwise, commit the whitespace!
++m_iterator;
if (auto result = try_commit_chunk(start_of_chunk, m_iterator, false, font, text_type); result.has_value())
m_current_index = next_grapheme_boundary();
if (auto result = try_commit_chunk(start_of_chunk, m_current_index, false, font, text_type); result.has_value())
return result.release_value();
continue;
}
}
++m_iterator;
m_current_index = next_grapheme_boundary();
}
if (start_of_chunk != m_utf8_view.end()) {
if (start_of_chunk != m_utf8_view.byte_length()) {
// Try to output whatever's left at the end of the text node.
if (auto result = try_commit_chunk(start_of_chunk, m_utf8_view.end(), false, font, text_type); result.has_value())
if (auto result = try_commit_chunk(start_of_chunk, m_utf8_view.byte_length(), false, font, text_type); result.has_value())
return result.release_value();
}
return {};
}
Optional<TextNode::Chunk> TextNode::ChunkIterator::try_commit_chunk(Utf8View::Iterator const& start, Utf8View::Iterator const& end, bool has_breaking_newline, Gfx::Font const& font, Gfx::GlyphRun::TextType text_type) const
Optional<TextNode::Chunk> TextNode::ChunkIterator::try_commit_chunk(size_t start, size_t end, bool has_breaking_newline, Gfx::Font const& font, Gfx::GlyphRun::TextType text_type) const
{
auto byte_offset = m_utf8_view.byte_offset_of(start);
auto byte_length = m_utf8_view.byte_offset_of(end) - byte_offset;
if (byte_length > 0) {
auto chunk_view = m_utf8_view.substring_view(byte_offset, byte_length);
if (auto byte_length = end - start; byte_length > 0) {
auto chunk_view = m_utf8_view.substring_view(start, byte_length);
return Chunk {
.view = chunk_view,
.font = font,
.start = byte_offset,
.start = start,
.length = byte_length,
.has_breaking_newline = has_breaking_newline,
.is_all_whitespace = is_all_whitespace(chunk_view.as_string()),