dmaax/ladybird
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

LibWeb: Verify whitespace font has the glyph in font_for_space()

Browse source 
When inline layout emits a whitespace chunk, it previously selected the
surrounding text's font without checking whether that font actually
contains a glyph for the whitespace codepoint. On pages that use
`@font-face` rules sharded by `unicode-range` (e.g. a Roboto webfont
split across one file for Cyrillic letters and another for basic Latin),
the shard covering the letters is picked for an adjacent space even
though the space codepoint lives in a different shard. HarfBuzz then
shapes the space with a font that has no glyph for it and emits
`.notdef`, rendering spaces as tofu boxes.

Check `contains_glyph(space_code_point)` on each candidate in
`font_for_space()` and fall through to
`FontCascadeList::font_for_code_point()` for the whitespace codepoint
when no surrounding font has the glyph.

Fixes whitespace rendering on web.telegram.org/a.
This commit is contained in:
Aliaksandr Kalenik 2026-04-22 16:43:53 +02:00 committed by Andreas Kling
parent 1f46651af5
commit 0bfe4677ae
5 changed files with 48 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

16
Libraries/LibWeb/Layout/TextNode.cpp
View file

@ -631,10 +631,12 @@ bool TextNode::ChunkIterator::is_at_line_break_opportunity() const
VERIFY_NOT_REACHED();
}
Gfx::Font const& TextNode::ChunkIterator::font_for_space(size_t at_index) const
Gfx::Font const& TextNode::ChunkIterator::font_for_space(size_t at_index, u32 space_code_point) const
{
auto has_glyph = [&](Gfx::Font const& font) { return font.contains_glyph(space_code_point); };
// 1. Prefer the last non-whitespace font in this node/run.
if (m_last_non_whitespace_font && !m_last_non_whitespace_font->is_emoji_font())
if (m_last_non_whitespace_font && !m_last_non_whitespace_font->is_emoji_font() && has_glyph(*m_last_non_whitespace_font))
return *m_last_non_whitespace_font;
// 2. Look ahead to the next non-space to infer the base font of this run.
@ -642,7 +644,7 @@ Gfx::Font const& TextNode::ChunkIterator::font_for_space(size_t at_index) const
auto cp = m_view.code_point_at(i);
if (!is_interword_space(cp) && cp != '\t' && cp != '\n') {
auto const& font = m_font_cascade_list.font_for_code_point(cp);
if (!font.is_emoji_font())
if (!font.is_emoji_font() && has_glyph(font))
return font;
// Text is coming from an emoji face; we'll fall back to (3).
break;
@ -650,8 +652,8 @@ Gfx::Font const& TextNode::ChunkIterator::font_for_space(size_t at_index) const
i = m_grapheme_segmenter.next_boundary(i).value_or(m_view.length_in_code_units());
}
// 3. No text around (leading/trailing/all spaces) — pick the first *text* face in the cascade.
return m_font_cascade_list.first_text_face();
// 3. No text around (leading/trailing/all spaces) — pick a font with the glyph from the cascade.
return m_font_cascade_list.font_for_code_point(space_code_point);
}
Optional<TextNode::Chunk> TextNode::ChunkIterator::next_without_peek()
@ -677,7 +679,7 @@ Optional<TextNode::Chunk> TextNode::ChunkIterator::next_without_peek()
auto const& expected_font_for = [&](u32 cp) -> Gfx::Font const& {
return is_interword_space(cp)
? font_for_space(m_current_index)
? font_for_space(m_current_index, cp)
: m_font_cascade_list.font_for_code_point(cp);
};
@ -749,7 +751,7 @@ Optional<TextNode::Chunk> TextNode::ChunkIterator::next_without_peek()
// Otherwise, commit the whitespace!
m_current_index = next_grapheme_boundary();
can_break_at_current_position = is_at_line_break_opportunity();
auto const& space_font = font_for_space(m_current_index);
auto const& space_font = font_for_space(m_current_index, code_point);
if (auto result = try_commit_chunk(start_of_chunk, m_current_index, false, broken_on_tab, false, space_font, text_type); result.has_value())
return result.release_value();
continue;

2
Libraries/LibWeb/Layout/TextNode.h
View file

@ -60,7 +60,7 @@ public:
Optional<Chunk> try_commit_chunk(size_t start, size_t end, bool has_breaking_newline, bool has_breaking_tab, bool can_break_after, Gfx::Font const&, Gfx::GlyphRun::TextType) const;
[[nodiscard]] bool is_at_line_break_opportunity() const;
[[nodiscard]] Gfx::Font const& font_for_space(size_t at_index) const;
[[nodiscard]] Gfx::Font const& font_for_space(size_t at_index, u32 space_code_point) const;
bool const m_should_wrap_lines;
bool const m_should_respect_linebreaks;

BIN
Tests/LibWeb/Assets/HashSansNoSpace.woff Normal file
View file

Binary file not shown.

1
Tests/LibWeb/Text/expected/font-cascade-space-fallback.txt Normal file
View file

@ -0,0 +1 @@
space matches fallback: true

37
Tests/LibWeb/Text/input/font-cascade-space-fallback.html Normal file
View file

@ -0,0 +1,37 @@
<!DOCTYPE html>
<style>
@font-face {
font-family: "HashFontNoSpace";
src: url("../../Assets/HashSansNoSpace.woff");
unicode-range: U+0041-005A;
}
.subset {
font-family: "HashFontNoSpace", "SerenitySans";
font-size: 100px;
}
.fallback-only {
font-family: "SerenitySans";
font-size: 100px;
}
span {
white-space: pre;
}
</style>
<script src="include.js"></script>
<div>
<span class="subset" id="letters-only">AB</span>
<span class="subset" id="with-space">A B</span>
<span class="fallback-only" id="ref-space">&nbsp;</span>
</div>
<script>
test(() => {
// HashFontNoSpace covers only A–Z; the space in "A B" must come from the
// SerenitySans fallback, not from HashFontNoSpace's .notdef glyph.
let lettersOnlyWidth = document.getElementById("letters-only").getBoundingClientRect().width;
let withSpaceWidth = document.getElementById("with-space").getBoundingClientRect().width;
let fallbackSpaceWidth = document.getElementById("ref-space").getBoundingClientRect().width;
let spaceWidth = withSpaceWidth - lettersOnlyWidth;
println("space matches fallback: " + (Math.abs(spaceWidth - fallbackSpaceWidth) < 2));
});
</script>