Match CSS text breaking semantics

This commit is contained in:
SGavrl committed 2026-08-03 11:32:45 +00:00
1 parent 34373c309e
commit 71193faddf
8 files changed
+1044 -51

No files matched your search

+253
View File
@@ -2128,6 +2128,8 @@ fn layout_dom_once(
list_style: crate::ListStyle, list_style: crate::ListStyle,
line_height: crate::LineHeight, line_height: crate::LineHeight,
white_space: crate::WhiteSpace, white_space: crate::WhiteSpace,
overflow_wrap: crate::OverflowWrap,
word_break: crate::WordBreak,
text_wrap_style: crate::TextWrapStyle, text_wrap_style: crate::TextWrapStyle,
text_transform: crate::TextTransform, text_transform: crate::TextTransform,
italic: bool, italic: bool,
@@ -2177,6 +2179,8 @@ fn layout_dom_once(
list_style: crate::ListStyle::Disc, list_style: crate::ListStyle::Disc,
line_height: crate::LineHeight::Normal, line_height: crate::LineHeight::Normal,
white_space: crate::WhiteSpace::Normal, white_space: crate::WhiteSpace::Normal,
overflow_wrap: crate::OverflowWrap::Normal,
word_break: crate::WordBreak::Normal,
text_wrap_style: crate::TextWrapStyle::Auto, text_wrap_style: crate::TextWrapStyle::Auto,
text_transform: crate::TextTransform::None, text_transform: crate::TextTransform::None,
italic: false, italic: false,
@@ -2587,6 +2591,8 @@ fn layout_dom_once(
match style.list_style { Some(v) => inh.list_style = v, None => style.list_style = Some(inh.list_style) } match style.list_style { Some(v) => inh.list_style = v, None => style.list_style = Some(inh.list_style) }
match style.line_height { Some(v) => inh.line_height = v, None => style.line_height = Some(inh.line_height) } match style.line_height { Some(v) => inh.line_height = v, None => style.line_height = Some(inh.line_height) }
match style.white_space { Some(v) => inh.white_space = v, None => style.white_space = Some(inh.white_space) } match style.white_space { Some(v) => inh.white_space = v, None => style.white_space = Some(inh.white_space) }
match style.overflow_wrap { Some(v) => inh.overflow_wrap = v, None => style.overflow_wrap = Some(inh.overflow_wrap) }
match style.word_break { Some(v) => inh.word_break = v, None => style.word_break = Some(inh.word_break) }
match style.text_wrap_style { Some(v) => inh.text_wrap_style = v, None => style.text_wrap_style = Some(inh.text_wrap_style) } match style.text_wrap_style { Some(v) => inh.text_wrap_style = v, None => style.text_wrap_style = Some(inh.text_wrap_style) }
match style.text_transform { Some(v) => inh.text_transform = v, None => style.text_transform = Some(inh.text_transform) } match style.text_transform { Some(v) => inh.text_transform = v, None => style.text_transform = Some(inh.text_transform) }
match style.font_style_italic { Some(v) => inh.italic = v, None => style.font_style_italic = Some(inh.italic) } match style.font_style_italic { Some(v) => inh.italic = v, None => style.font_style_italic = Some(inh.italic) }
@@ -2708,6 +2714,8 @@ fn layout_dom_once(
let host_variation_settings = style.font_variation_settings.clone(); let host_variation_settings = style.font_variation_settings.clone();
let host_line_height = style.line_height; let host_line_height = style.line_height;
let host_white_space = style.white_space; let host_white_space = style.white_space;
let host_overflow_wrap = style.overflow_wrap;
let host_word_break = style.word_break;
let host_text_wrap_style = style.text_wrap_style; let host_text_wrap_style = style.text_wrap_style;
let host_transform = style.text_transform; let host_transform = style.text_transform;
let host_italic = style.font_style_italic; let host_italic = style.font_style_italic;
@@ -2935,6 +2943,12 @@ fn layout_dom_once(
if pseudo.white_space.is_none() { if pseudo.white_space.is_none() {
pseudo.white_space = host_white_space; pseudo.white_space = host_white_space;
} }
if pseudo.overflow_wrap.is_none() {
pseudo.overflow_wrap = host_overflow_wrap;
}
if pseudo.word_break.is_none() {
pseudo.word_break = host_word_break;
}
if pseudo.text_wrap_style.is_none() { if pseudo.text_wrap_style.is_none() {
pseudo.text_wrap_style = host_text_wrap_style; pseudo.text_wrap_style = host_text_wrap_style;
} }
@@ -11760,4 +11774,243 @@ mod tests {
assert!(!laid.styles[&grid].grid_auto_rows_inherit); assert!(!laid.styles[&grid].grid_auto_rows_inherit);
} }
#[cfg(feature = "paint")]
#[test]
fn text_break_modes_match_chromium_fixed_width_geometry() {
// Chromium 145 at DPR 1 with Liberation Sans produces the same
// 20/40px line boxes. `break-word` and `anywhere` are emergency
// wrapping modes; `break-all` adds typographic-letter opportunities.
// keep-all must retain ordinary Latin word/space behavior.
let tree = parse_html(
r#"<style>
html,body { margin:0 }
.fixed { width:80px;font:16px/20px "Liberation Sans" }
#normal { overflow-wrap:normal }
#break-word { overflow-wrap:break-word }
#anywhere { overflow-wrap:anywhere }
#break-all { word-break:break-all }
#keep-all { word-break:keep-all }
#legacy { word-break:break-word }
</style>
<div id="normal" class="fixed">abcdefghijklmnop</div>
<div id="break-word" class="fixed">abcdefghijklmnop</div>
<div id="anywhere" class="fixed">abcdefghijklmnop</div>
<div id="break-all" class="fixed">abcdefghijklmnop</div>
<div id="latin-normal" class="fixed">alpha beta gamma</div>
<div id="keep-all" class="fixed">alpha beta gamma</div>
<div id="legacy" class="fixed">abcdefghijklmnop</div>"#,
);
let laid = layout_dom(&tree, (500.0, 500.0));
let rect = |id| laid.rects[&tree.get_element_by_id(id).unwrap()];
assert_eq!(
rect("normal"),
Rect { x: 0.0, y: 0.0, width: 80.0, height: 20.0 }
);
assert_eq!(
rect("break-word"),
Rect { x: 0.0, y: 20.0, width: 80.0, height: 40.0 }
);
assert_eq!(
rect("anywhere"),
Rect { x: 0.0, y: 60.0, width: 80.0, height: 40.0 }
);
assert_eq!(
rect("break-all"),
Rect { x: 0.0, y: 100.0, width: 80.0, height: 40.0 }
);
assert_eq!(rect("latin-normal").height, 40.0);
assert_eq!(rect("keep-all").height, rect("latin-normal").height);
assert_eq!(rect("legacy").height, 40.0);
}
#[cfg(feature = "paint")]
#[test]
fn text_break_inheritance_alias_and_css_wide_values_match_chromium() {
let tree = parse_html(
r#"<style>
html,body { margin:0 }
.parent { overflow-wrap:anywhere;word-break:break-all;font:16px/20px "Liberation Sans" }
.fixed { width:80px }
#initial { overflow-wrap:initial;word-break:initial }
#unset { overflow-wrap:unset;word-break:unset }
#inherit { overflow-wrap:inherit;word-break:inherit }
#revert { overflow-wrap:revert;word-break:revert }
#revert-layer { overflow-wrap:revert-layer;word-break:revert-layer }
#alias { word-wrap:anywhere;word-break:normal }
#inherited::before { content:"";display:none }
</style>
<section class="parent">
<div id="inherited" class="fixed">abcdefghijklmnop</div>
<div id="initial" class="fixed">abcdefghijklmnop</div>
<div id="unset" class="fixed">abcdefghijklmnop</div>
<div id="inherit" class="fixed">abcdefghijklmnop</div>
<div id="revert" class="fixed">abcdefghijklmnop</div>
<div id="revert-layer" class="fixed">abcdefghijklmnop</div>
<div id="alias" class="fixed">abcdefghijklmnop</div>
</section>"#,
);
let laid = layout_dom(&tree, (500.0, 500.0));
let node = |id| tree.get_element_by_id(id).unwrap();
let style = |id| &laid.styles[&node(id)];
let height = |id| laid.rects[&node(id)].height;
assert_eq!(style("inherited").overflow_wrap, Some(crate::OverflowWrap::Anywhere));
assert_eq!(style("inherited").word_break, Some(crate::WordBreak::BreakAll));
assert_eq!(height("inherited"), 40.0);
assert_eq!(
style("inherited").before_pseudo.as_ref().unwrap().overflow_wrap,
Some(crate::OverflowWrap::Anywhere)
);
assert_eq!(
style("inherited").before_pseudo.as_ref().unwrap().word_break,
Some(crate::WordBreak::BreakAll)
);
assert_eq!(style("initial").overflow_wrap, Some(crate::OverflowWrap::Normal));
assert_eq!(style("initial").word_break, Some(crate::WordBreak::Normal));
assert_eq!(height("initial"), 20.0);
for id in ["unset", "inherit", "revert", "revert-layer"] {
assert_eq!(style(id).overflow_wrap, Some(crate::OverflowWrap::Anywhere), "{id}");
assert_eq!(style(id).word_break, Some(crate::WordBreak::BreakAll), "{id}");
assert_eq!(height(id), 40.0, "{id}");
}
assert_eq!(style("alias").overflow_wrap, Some(crate::OverflowWrap::Anywhere));
assert_eq!(style("alias").word_break, Some(crate::WordBreak::Normal));
assert_eq!(height("alias"), 40.0);
}
#[cfg(feature = "paint")]
#[test]
fn anywhere_but_not_break_word_reduces_min_content() {
let tree = parse_html(
r#"<style>
html,body { margin:0 }
.grid { display:grid;grid-template-columns:min-content;font:16px/20px "Liberation Sans" }
.cell { display:block }
#normal { overflow-wrap:normal }
#break-word { overflow-wrap:break-word }
#anywhere { overflow-wrap:anywhere }
#break-all { word-break:break-all }
</style>
<div class="grid"><div id="normal" class="cell">abcdefghijklmnop</div></div>
<div class="grid"><div id="break-word" class="cell">abcdefghijklmnop</div></div>
<div class="grid"><div id="anywhere" class="cell">abcdefghijklmnop</div></div>
<div class="grid"><div id="break-all" class="cell">abcdefghijklmnop</div></div>"#,
);
let laid = layout_dom(&tree, (500.0, 1000.0));
let width = |id: &str| laid.rects[&tree.get_element_by_id(id).unwrap()].width;
assert_eq!(width("normal"), 125.0);
assert_eq!(width("break-word"), 125.0);
assert_eq!(width("anywhere"), 14.0);
assert_eq!(width("break-all"), 14.0);
}
#[cfg(feature = "paint")]
#[test]
fn inline_descendant_break_policies_stay_at_their_style_boundaries() {
let tree = parse_html(
r#"<style>
html,body { margin:0 }
.fixed { width:80px;font:16px/20px "Liberation Sans" }
.normal { word-break:normal;overflow-wrap:normal }
.break-all { word-break:break-all }
.anywhere { overflow-wrap:anywhere }
</style>
<div id="child-break-all" class="fixed normal"><span class="break-all">abcdefghijklmnop</span></div>
<div id="child-normal" class="fixed break-all"><span class="normal">abcdefghijklmnop</span></div>
<div id="child-anywhere" class="fixed normal"><span class="anywhere">abcdefghijklmnop</span></div>"#,
);
let laid = layout_dom(&tree, (500.0, 300.0));
let height = |id| laid.rects[&tree.get_element_by_id(id).unwrap()].height;
assert_eq!(height("child-break-all"), 40.0);
assert_eq!(height("child-normal"), 20.0);
assert_eq!(height("child-anywhere"), 40.0);
}
#[cfg(feature = "paint")]
#[test]
fn break_all_keeps_uax14_punctuation_pairs_in_min_content() {
let tree = parse_html(
r#"<style>
html,body { margin:0 }
.grid { display:grid;grid-template-columns:min-content;font:16px/20px "Liberation Sans" }
#break-all { word-break:break-all }
#anywhere { overflow-wrap:anywhere }
</style>
<div class="grid"><span id="break-all">(ab)</span></div>
<div class="grid"><span id="anywhere">(ab)</span></div>"#,
);
let laid = layout_dom(&tree, (500.0, 300.0));
let rect = |id| laid.rects[&tree.get_element_by_id(id).unwrap()];
assert!(rect("break-all").width > rect("anywhere").width);
assert_eq!(rect("break-all").height, 40.0);
assert_eq!(rect("anywhere").height, 80.0);
}
#[cfg(feature = "paint")]
#[test]
fn break_all_uses_blink_punctuation_pair_semantics() {
let tree = parse_html(
r#"<style>
html,body { margin:0 }
.grid { display:grid;grid-template-columns:min-content;font:20px/24px monospace }
.break { word-break:break-all }
</style>
<div class="grid"><span id="close-normal">)a</span></div>
<div class="grid"><span id="close-break" class="break">)a</span></div>
<div class="grid"><span id="closing-normal">a)</span></div>
<div class="grid"><span id="closing-break" class="break">a)</span></div>
<div class="grid"><span id="slash-normal">a/b</span></div>
<div class="grid"><span id="slash-break" class="break">a/b</span></div>
<div class="grid"><span id="hyphen-normal">a-b</span></div>
<div class="grid"><span id="hyphen-break" class="break">a-b</span></div>
<div class="grid"><span id="bar-normal">a|b</span></div>
<div class="grid"><span id="bar-break" class="break">a|b</span></div>
<div class="grid"><span id="plus-normal">a+b</span></div>
<div class="grid"><span id="plus-break" class="break">a+b</span></div>
<div class="grid"><span id="prefix-normal">$a</span></div>
<div class="grid"><span id="prefix-break" class="break">$a</span></div>
<div class="grid"><span id="postfix-normal">a%</span></div>
<div class="grid"><span id="postfix-break" class="break">a%</span></div>"#,
);
let laid = layout_dom(&tree, (500.0, 1000.0));
let width = |id: &str| laid.rects[&tree.get_element_by_id(id).unwrap()].width;
for stem in ["close", "slash", "hyphen", "bar", "plus"] {
assert!(
width(&format!("{stem}-break")) < width(&format!("{stem}-normal")),
"break-all should expose a narrower opportunity for {stem}"
);
}
for stem in ["closing", "prefix", "postfix"] {
assert_eq!(
width(&format!("{stem}-break")),
width(&format!("{stem}-normal")),
"break-all must retain the prohibited boundary for {stem}"
);
}
}
#[cfg(feature = "paint")]
#[test]
fn keep_all_suppresses_mixed_cjk_latin_word_boundaries() {
let tree = parse_html(
r#"<style>
html,body { margin:0 }
.grid { display:grid;grid-template-columns:min-content;font:16px/20px "Liberation Sans" }
#keep { word-break:keep-all }
</style>
<div class="grid"><span id="normal">標(準)萬ab國.碼</span></div>
<div class="grid"><span id="keep">標(準)萬ab國.碼</span></div>"#,
);
let laid = layout_dom(&tree, (500.0, 500.0));
let width = |id| laid.rects[&tree.get_element_by_id(id).unwrap()].width;
assert!(width("keep") > width("normal"));
}
} }
+148 -23
View File
@@ -15,9 +15,9 @@
use std::{collections::HashMap, sync::Arc}; use std::{collections::HashMap, sync::Arc};
use cosmic_text::{ use cosmic_text::{
Align, Attrs, Buffer, CacheKey, CacheKeyFlags, Color, Family, FeatureTag, FontFeatures, Align, Attrs, Buffer, CacheKey, CacheKeyFlags, Color, CssLineBreak, CssOverflowWrap,
FontSystem, FontVariations, Metrics, Shaping, Style, SwashCache, SwashImage, VariationTag, CssWordBreak, Family, FeatureTag, FontFeatures, FontSystem, FontVariations, Metrics, Shaping,
Weight, Wrap, Style, SwashCache, SwashImage, VariationTag, Weight, Wrap,
}; };
use swash::scale::{image::Content as SwashContent, Render, ScaleContext, Source, StrikeWith}; use swash::scale::{image::Content as SwashContent, Render, ScaleContext, Source, StrikeWith};
use swash::zeno::{Angle, Format, Transform, Vector}; use swash::zeno::{Angle, Format, Transform, Vector};
@@ -381,6 +381,12 @@ fn metadata_variation(metadata: usize) -> Option<usize> {
/// paint it (filled in after layout). /// paint it (filled in after layout).
pub struct InlineItem { pub struct InlineItem {
buffer: Buffer, buffer: Buffer,
/// Wrapping used for definite-width layout and final paint.
layout_wrap: Wrap,
/// Wrapping used only for an intrinsic min-content query. This differs
/// from `layout_wrap` for `overflow-wrap: break-word`: emergency breaks
/// are available during reflow but do not reduce min-content.
min_content_wrap: Wrap,
/// Unmodified shaped content retained only when `text-indent` is nonzero. /// Unmodified shaped content retained only when `text-indent` is nonzero.
/// Each Taffy measurement can then derive a first-line-only wrap boundary /// Each Taffy measurement can then derive a first-line-only wrap boundary
/// without accumulating synthetic hard breaks across repeated probes. /// without accumulating synthetic hard breaks across repeated probes.
@@ -1019,6 +1025,9 @@ impl TextEngine {
color: context.color, color: context.color,
family: context.family, family: context.family,
clip_fill: context.clip_fill, clip_fill: context.clip_fill,
white_space: context.white_space,
overflow_wrap: context.overflow_wrap,
word_break: context.word_break,
}; };
let mut spans = Vec::new(); let mut spans = Vec::new();
push_text( push_text(
@@ -1043,6 +1052,19 @@ impl TextEngine {
clip_fills: Vec<ClipTextFill>, clip_fills: Vec<ClipTextFill>,
) -> Option<usize> { ) -> Option<usize> {
let white_space = base.white_space.unwrap_or_default(); let white_space = base.white_space.unwrap_or_default();
let layout_wrap = if spans.iter().any(|(_, attrs)| attrs.has_layout_emergency_breaks()) {
Wrap::WordOrGlyph
} else {
Wrap::Word
};
let min_content_wrap = if spans
.iter()
.any(|(_, attrs)| attrs.has_min_content_emergency_breaks())
{
Wrap::WordOrGlyphMinContent
} else {
Wrap::Word
};
// Collapsible trailing whitespace does not widen the last line. // Collapsible trailing whitespace does not widen the last line.
if matches!( if matches!(
white_space, white_space,
@@ -1095,22 +1117,12 @@ impl TextEngine {
let cosmic_size = base_size.max(1.0); let cosmic_size = base_size.max(1.0);
let metrics = Metrics::new(cosmic_size, line_h.max(1.0)); let metrics = Metrics::new(cosmic_size, line_h.max(1.0));
let mut buffer = Buffer::new(&mut self.font_system, metrics); let mut buffer = Buffer::new(&mut self.font_system, metrics);
// Break only at word boundaries, never mid-word. This keeps the // Install the used-layout mode now; intrinsic measurement temporarily
// min-content width (measured at width 0) equal to the longest word, // swaps to `min_content_wrap` below. Keeping those modes separate is
// so a table/flex column never squeezes narrower than a whole word // what prevents `overflow-wrap: break-word` from shrinking a
// and forces an ugly mid-word break; a single word wider than its // table/flex min-content contribution while still permitting an
// column overflows instead, matching normal `overflow-wrap: normal`. // emergency break once the actual line width is known.
buffer.set_wrap( buffer.set_wrap(&mut self.font_system, layout_wrap);
&mut self.font_system,
if matches!(
white_space,
crate::WhiteSpace::NoWrap | crate::WhiteSpace::Pre
) {
Wrap::None
} else {
Wrap::Word
},
);
// Always Advanced shaping: Basic mis-maps per-span attribute // Always Advanced shaping: Basic mis-maps per-span attribute
// boundaries in the shaping backend (a multi-color run like body text // boundaries in the shaping backend (a multi-color run like body text
// with links ends up coloring the wrong glyphs), and shaping is a // with links ends up coloring the wrong glyphs), and shaping is a
@@ -1166,6 +1178,8 @@ impl TextEngine {
.then(|| buffer.clone()); .then(|| buffer.clone());
self.items.push(InlineItem { self.items.push(InlineItem {
buffer, buffer,
layout_wrap,
min_content_wrap,
source_buffer, source_buffer,
text_indent, text_indent,
first_line_offset: 0.0, first_line_offset: 0.0,
@@ -1189,9 +1203,19 @@ impl TextEngine {
.size(taffy::Size { width, height: None }); .size(taffy::Size { width, height: None });
return (size.width, size.height); return (size.width, size.height);
} }
let wrap = self.items[idx].layout_wrap;
self.measure_text_with_wrap(idx, width, wrap)
}
fn measure_text_with_wrap(
&mut self,
idx: usize,
width: Option<f32>,
wrap: Wrap,
) -> (f32, f32) {
let TextEngine { font_system, items, .. } = self; let TextEngine { font_system, items, .. } = self;
let item = &mut items[idx]; let item = &mut items[idx];
shape_with_text_indent(font_system, item, width); shape_with_text_indent(font_system, item, width, wrap);
let (width, height) = buffer_size(&item.buffer, item.first_line_offset); let (width, height) = buffer_size(&item.buffer, item.first_line_offset);
(width, height.max(item.forced_min_height)) (width, height.max(item.forced_min_height))
} }
@@ -1207,7 +1231,8 @@ impl TextEngine {
let Some(item) = items.get_mut(idx) else { let Some(item) = items.get_mut(idx) else {
return (0.0, 0.0); return (0.0, 0.0);
}; };
shape_with_text_indent(font_system, item, None); let wrap = item.layout_wrap;
shape_with_text_indent(font_system, item, None, wrap);
let mut width = 0.0f32; let mut width = 0.0f32;
let mut height = 0.0f32; let mut height = 0.0f32;
for (line_index, run) in item.buffer.layout_runs().enumerate() { for (line_index, run) in item.buffer.layout_runs().enumerate() {
@@ -1257,12 +1282,19 @@ impl TextEngine {
} }
return size; return size;
} }
let min_content_query =
known.width.is_none() && matches!(available.width, taffy::AvailableSpace::MinContent);
let width = known.width.or(match available.width { let width = known.width.or(match available.width {
taffy::AvailableSpace::Definite(width) => Some(width), taffy::AvailableSpace::Definite(width) => Some(width),
taffy::AvailableSpace::MinContent => Some(0.0), taffy::AvailableSpace::MinContent => Some(0.0),
taffy::AvailableSpace::MaxContent => None, taffy::AvailableSpace::MaxContent => None,
}); });
let (width, height) = self.measure(idx, width); let wrap = if min_content_query {
self.items[idx].min_content_wrap
} else {
self.items[idx].layout_wrap
};
let (width, height) = self.measure_text_with_wrap(idx, width, wrap);
taffy::Size { width, height } taffy::Size { width, height }
} }
@@ -1281,7 +1313,8 @@ impl TextEngine {
} = self; } = self;
let item = &mut items[idx]; let item = &mut items[idx];
let content_width = content_width.max(0.0); let content_width = content_width.max(0.0);
shape_with_text_indent(font_system, item, Some(content_width)); let wrap = item.layout_wrap;
shape_with_text_indent(font_system, item, Some(content_width), wrap);
let balanced_width = item let balanced_width = item
.balance_wrap .balance_wrap
.then(|| balance_wrap_width(font_system, &mut item.buffer, content_width)) .then(|| balance_wrap_width(font_system, &mut item.buffer, content_width))
@@ -1315,9 +1348,31 @@ struct SpanAttrs {
color: [u8; 4], color: [u8; 4],
family: Arc<str>, family: Arc<str>,
clip_fill: Option<usize>, clip_fill: Option<usize>,
white_space: crate::WhiteSpace,
overflow_wrap: crate::OverflowWrap,
word_break: crate::WordBreak,
} }
impl SpanAttrs { impl SpanAttrs {
fn wrapping_enabled(&self) -> bool {
!matches!(
self.white_space,
crate::WhiteSpace::NoWrap | crate::WhiteSpace::Pre
)
}
fn has_layout_emergency_breaks(&self) -> bool {
self.wrapping_enabled()
&& (self.word_break == crate::WordBreak::BreakWord
|| self.overflow_wrap != crate::OverflowWrap::Normal)
}
fn has_min_content_emergency_breaks(&self) -> bool {
self.wrapping_enabled()
&& (self.word_break == crate::WordBreak::BreakWord
|| self.overflow_wrap == crate::OverflowWrap::Anywhere)
}
fn to_attrs(&self, variation_index: usize) -> Attrs<'_> { fn to_attrs(&self, variation_index: usize) -> Attrs<'_> {
let mut a = Attrs::new().family(Family::Name(self.family.as_ref())); let mut a = Attrs::new().family(Family::Name(self.family.as_ref()));
// Inline descendants keep their own computed font metrics inside the // Inline descendants keep their own computed font metrics inside the
@@ -1371,6 +1426,22 @@ impl SpanAttrs {
if let Some(variations) = self.variations.as_ref() { if let Some(variations) = self.variations.as_ref() {
a = a.font_variations((**variations).clone()); a = a.font_variations((**variations).clone());
} }
let word_break = match self.word_break {
crate::WordBreak::Normal => CssWordBreak::Normal,
crate::WordBreak::BreakAll => CssWordBreak::BreakAll,
crate::WordBreak::KeepAll => CssWordBreak::KeepAll,
crate::WordBreak::BreakWord => CssWordBreak::BreakWord,
};
let overflow_wrap = match self.overflow_wrap {
crate::OverflowWrap::Normal => CssOverflowWrap::Normal,
crate::OverflowWrap::BreakWord => CssOverflowWrap::BreakWord,
crate::OverflowWrap::Anywhere => CssOverflowWrap::Anywhere,
};
a = a.css_line_break(CssLineBreak {
wrap: self.wrapping_enabled(),
word_break,
overflow_wrap,
});
a = a.metadata(fill | variation | usize::from(self.underline)); a = a.metadata(fill | variation | usize::from(self.underline));
a a
} }
@@ -1394,6 +1465,8 @@ struct SpanCtx {
underline: bool, underline: bool,
transform: TextTransform, transform: TextTransform,
white_space: crate::WhiteSpace, white_space: crate::WhiteSpace,
overflow_wrap: crate::OverflowWrap,
word_break: crate::WordBreak,
family: Arc<str>, family: Arc<str>,
clip_fill: Option<usize>, clip_fill: Option<usize>,
} }
@@ -1460,6 +1533,8 @@ fn base_span_ctx(
underline: base.underline.unwrap_or(false), underline: base.underline.unwrap_or(false),
transform: base.text_transform.unwrap_or(TextTransform::None), transform: base.text_transform.unwrap_or(TextTransform::None),
white_space: base.white_space.unwrap_or_default(), white_space: base.white_space.unwrap_or_default(),
overflow_wrap: base.overflow_wrap.unwrap_or_default(),
word_break: base.word_break.unwrap_or_default(),
family: font.family, family: font.family,
clip_fill, clip_fill,
} }
@@ -1518,6 +1593,9 @@ fn collect_node_spans(
color: ctx.color, color: ctx.color,
family: Arc::clone(&ctx.family), family: Arc::clone(&ctx.family),
clip_fill: ctx.clip_fill, clip_fill: ctx.clip_fill,
white_space: ctx.white_space,
overflow_wrap: ctx.overflow_wrap,
word_break: ctx.word_break,
}; };
push_text( push_text(
contents, contents,
@@ -1552,6 +1630,9 @@ fn collect_node_spans(
color: ctx.color, color: ctx.color,
family: Arc::clone(&ctx.family), family: Arc::clone(&ctx.family),
clip_fill: ctx.clip_fill, clip_fill: ctx.clip_fill,
white_space: ctx.white_space,
overflow_wrap: ctx.overflow_wrap,
word_break: ctx.word_break,
}, },
)); ));
c.last_was_space = true; c.last_was_space = true;
@@ -1623,6 +1704,10 @@ fn collect_node_spans(
white_space: style white_space: style
.and_then(|s| s.white_space) .and_then(|s| s.white_space)
.unwrap_or(ctx.white_space), .unwrap_or(ctx.white_space),
overflow_wrap: style
.and_then(|s| s.overflow_wrap)
.unwrap_or(ctx.overflow_wrap),
word_break: style.and_then(|s| s.word_break).unwrap_or(ctx.word_break),
family: font.family, family: font.family,
clip_fill, clip_fill,
}; };
@@ -1725,6 +1810,7 @@ fn shape_with_text_indent(
font_system: &mut FontSystem, font_system: &mut FontSystem,
item: &mut InlineItem, item: &mut InlineItem,
width: Option<f32>, width: Option<f32>,
wrap: Wrap,
) { ) {
let indent = used_text_indent(item.text_indent, width); let indent = used_text_indent(item.text_indent, width);
item.first_line_offset = indent item.first_line_offset = indent
@@ -1735,12 +1821,14 @@ fn shape_with_text_indent(
}; };
let Some(source) = item.source_buffer.as_ref() else { let Some(source) = item.source_buffer.as_ref() else {
item.buffer.set_wrap(font_system, wrap);
item.buffer item.buffer
.set_size(font_system, width.map(|value| value.max(0.0)), None); .set_size(font_system, width.map(|value| value.max(0.0)), None);
item.buffer.shape_until_scroll(font_system, false); item.buffer.shape_until_scroll(font_system, false);
return; return;
}; };
item.buffer = source.clone(); item.buffer = source.clone();
item.buffer.set_wrap(font_system, wrap);
let mut first_break = None; let mut first_break = None;
if let Some(full_width) = width { if let Some(full_width) = width {
@@ -1755,6 +1843,7 @@ fn shape_with_text_indent(
} }
item.buffer = source.clone(); item.buffer = source.clone();
item.buffer.set_wrap(font_system, wrap);
if let (Some(mut split), Some(first_line)) = if let (Some(mut split), Some(first_line)) =
(first_break, item.buffer.lines.first()) (first_break, item.buffer.lines.first())
{ {
@@ -2758,6 +2847,9 @@ mod tests {
color: [1, 2, 3, 255], color: [1, 2, 3, 255],
family: Arc::from(FAMILY), family: Arc::from(FAMILY),
clip_fill: Some(37), clip_fill: Some(37),
white_space: crate::WhiteSpace::Normal,
overflow_wrap: crate::OverflowWrap::Normal,
word_break: crate::WordBreak::Normal,
}; };
let shaped = attrs.to_attrs(42); let shaped = attrs.to_attrs(42);
assert_ne!(shaped.metadata & META_UNDERLINE, 0); assert_ne!(shaped.metadata & META_UNDERLINE, 0);
@@ -2835,6 +2927,9 @@ mod tests {
color: [0, 0, 0, 255], color: [0, 0, 0, 255],
family: Arc::from(FAMILY), family: Arc::from(FAMILY),
clip_fill: None, clip_fill: None,
white_space: crate::WhiteSpace::Normal,
overflow_wrap: crate::OverflowWrap::Normal,
word_break: crate::WordBreak::Normal,
}; };
let attrs = span.to_attrs(1); let attrs = span.to_attrs(1);
assert_eq!(attrs.letter_spacing_opt.unwrap().0, 0.1); assert_eq!(attrs.letter_spacing_opt.unwrap().0, 0.1);
@@ -2850,6 +2945,36 @@ mod tests {
assert!(zero.to_attrs(1).font_features.features.is_empty()); assert!(zero.to_attrs(1).font_features.features.is_empty());
} }
#[test]
fn keep_all_never_inserts_controls_inside_grapheme_clusters() {
let source = "각각 か\u{3099}か\u{3099} 👨‍👩‍👧‍👦 👍🏽";
let tree = obscura_dom::parse_html(&format!("<p id='copy'>{source}</p>"));
let copy = tree.get_element_by_id("copy").unwrap();
let base = LayoutStyle {
display: Display::Block,
font_size: Some(40.0),
line_height: Some(crate::LineHeight::Px(48.0)),
..Default::default()
};
let mut keep = base.clone();
keep.word_break = Some(crate::WordBreak::KeepAll);
let mut normal_engine = TextEngine::new();
let normal_item = normal_engine
.try_build(&tree, copy, &HashMap::from([(copy, base)]))
.unwrap();
let normal_size = normal_engine.measure(normal_item, None);
let mut keep_engine = TextEngine::new();
let keep_item = keep_engine
.try_build(&tree, copy, &HashMap::from([(copy, keep)]))
.unwrap();
let keep_size = keep_engine.measure(keep_item, None);
assert_eq!(keep_engine.item_text(keep_item), source);
assert_eq!(keep_size, normal_size);
}
fn variable_font_fixture() -> Vec<u8> { fn variable_font_fixture() -> Vec<u8> {
let encoded: String = include_str!("../tests/fonts/obscura-vf-test.woff2.b64") let encoded: String = include_str!("../tests/fonts/obscura-vf-test.woff2.b64")
.chars() .chars()
+34
View File
@@ -945,6 +945,19 @@ pub struct LayoutStyle {
/// value (or the initial `normal`). The inline shaper uses this to retain /// value (or the initial `normal`). The inline shaper uses this to retain
/// author/source newlines in code blocks and to select wrapping behavior. /// author/source newlines in code blocks and to select wrapping behavior.
pub white_space: Option<WhiteSpace>, pub white_space: Option<WhiteSpace>,
/// `overflow-wrap` (legacy alias: `word-wrap`). Inherited; `None` means
/// inherit the nearest ancestor's value (or the initial `normal`).
///
/// `Anywhere` contributes its emergency grapheme opportunities to
/// min-content sizing, while legacy `BreakWord` uses those opportunities
/// only for actual line layout.
pub overflow_wrap: Option<OverflowWrap>,
/// `word-break`. Inherited; `None` means inherit the nearest ancestor's
/// value (or the initial `normal`). Kept separate from `overflow-wrap`
/// because `break-all` adds typographic-letter opportunities while
/// retaining UAX#14 punctuation constraints, whereas `keep-all`
/// suppresses eligible letter/number boundaries without altering text.
pub word_break: Option<WordBreak>,
/// `text-wrap-style`. Inherited; `None` means inherit the nearest /// `text-wrap-style`. Inherited; `None` means inherit the nearest
/// ancestor's value (or the initial `auto`). `Balance` keeps the natural /// ancestor's value (or the initial `auto`). `Balance` keeps the natural
/// line count but tightens the effective line-breaking width during the /// line count but tightens the effective line-breaking width during the
@@ -1166,6 +1179,27 @@ pub enum WhiteSpace {
BreakSpaces, BreakSpaces,
} }
/// Emergency wrapping behavior for otherwise-unbreakable text.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
pub enum OverflowWrap {
#[default]
Normal,
BreakWord,
Anywhere,
}
/// The supported `word-break` values. `BreakWord` is the legacy compatibility
/// value whose effective behavior is `word-break: normal` plus
/// `overflow-wrap: anywhere`, matching Blink and Gecko.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
pub enum WordBreak {
#[default]
Normal,
BreakAll,
KeepAll,
BreakWord,
}
/// Deterministic CSS animation sample time relative to the animation's /// Deterministic CSS animation sample time relative to the animation's
/// timeline origin. Screenshot rendering currently uses the initial value /// timeline origin. Screenshot rendering currently uses the initial value
/// (0ms); keeping it explicit lets capture plumbing freeze other engines at /// (0ms); keeping it explicit lets capture plumbing freeze other engines at
+36
View File
@@ -501,6 +501,25 @@ impl PreparedRender {
} }
.to_string(), .to_string(),
); );
let overflow_wrap = match style.overflow_wrap.unwrap_or_default() {
crate::OverflowWrap::Normal => "normal",
crate::OverflowWrap::BreakWord => "break-word",
crate::OverflowWrap::Anywhere => "anywhere",
}
.to_string();
out.insert("overflow-wrap", overflow_wrap.clone());
// CSSOM retains the legacy alias as a separately addressable property.
out.insert("word-wrap", overflow_wrap);
out.insert(
"word-break",
match style.word_break.unwrap_or_default() {
crate::WordBreak::Normal => "normal",
crate::WordBreak::BreakAll => "break-all",
crate::WordBreak::KeepAll => "keep-all",
crate::WordBreak::BreakWord => "break-word",
}
.to_string(),
);
out.insert( out.insert(
"text-align", "text-align",
match style.text_align { match style.text_align {
@@ -7886,6 +7905,23 @@ mod tests {
assert!(found, "computed color should resolve currentColor in inline svg"); assert!(found, "computed color should resolve currentColor in inline svg");
} }
#[test]
fn computed_style_exposes_text_break_longhands_and_alias() {
let tree = parse_html(
r#"<style>#copy{overflow-wrap:anywhere;word-break:keep-all}</style>
<p id="copy">copy</p>"#,
);
let copy = tree.get_element_by_id("copy").unwrap();
let mut resources = RenderResourceCache::default();
let prepared = prepare_dom(&tree, (320.0, 200.0), None, &mut resources)
.expect("prepared render");
let computed = prepared.computed_style(copy).expect("computed style");
assert_eq!(computed.get("overflow-wrap").map(String::as_str), Some("anywhere"));
assert_eq!(computed.get("word-wrap").map(String::as_str), Some("anywhere"));
assert_eq!(computed.get("word-break").map(String::as_str), Some("keep-all"));
}
#[test] #[test]
fn paints_inline_svg_with_framework_colon_attribute() { fn paints_inline_svg_with_framework_colon_attribute() {
let tree = obscura_dom::parse_html( let tree = obscura_dom::parse_html(
+82
View File
@@ -1454,6 +1454,28 @@ fn apply_value(style: &mut LayoutStyle, name: &str, value: &str) {
_ => style.white_space, _ => style.white_space,
}; };
} }
"overflow-wrap" | "word-wrap" => {
style.overflow_wrap = match value.trim().to_ascii_lowercase().as_str() {
"normal" | "initial" => Some(crate::OverflowWrap::Normal),
"break-word" => Some(crate::OverflowWrap::BreakWord),
"anywhere" => Some(crate::OverflowWrap::Anywhere),
// Both properties inherit. With no lower-origin declaration
// in the compact cascade, revert also exposes the inherited
// value, as Blink and Gecko do for these inherited longhands.
"inherit" | "unset" | "revert" | "revert-layer" => None,
_ => style.overflow_wrap,
};
}
"word-break" => {
style.word_break = match value.trim().to_ascii_lowercase().as_str() {
"normal" | "initial" => Some(crate::WordBreak::Normal),
"break-all" => Some(crate::WordBreak::BreakAll),
"keep-all" => Some(crate::WordBreak::KeepAll),
"break-word" => Some(crate::WordBreak::BreakWord),
"inherit" | "unset" | "revert" | "revert-layer" => None,
_ => style.word_break,
};
}
"text-wrap" => { "text-wrap" => {
style.text_wrap_style = match value.trim().to_ascii_lowercase().as_str() { style.text_wrap_style = match value.trim().to_ascii_lowercase().as_str() {
"auto" | "wrap" | "initial" | "revert" | "revert-layer" => { "auto" | "wrap" | "initial" | "revert" | "revert-layer" => {
@@ -1717,6 +1739,9 @@ pub(crate) fn supports_declaration(name: &str, value: &str) -> bool {
| "text-decoration-line" | "text-decoration-line"
| "line-height" | "line-height"
| "white-space" | "white-space"
| "overflow-wrap"
| "word-wrap"
| "word-break"
| "text-wrap" | "text-wrap"
| "text-wrap-style" | "text-wrap-style"
| "align-items" | "align-items"
@@ -1858,6 +1883,14 @@ pub(crate) fn supports_declaration(name: &str, value: &str) -> bool {
value.to_ascii_lowercase().as_str(), value.to_ascii_lowercase().as_str(),
"normal" | "nowrap" | "pre" | "pre-wrap" | "pre-line" | "break-spaces" "normal" | "nowrap" | "pre" | "pre-wrap" | "pre-line" | "break-spaces"
), ),
"overflow-wrap" | "word-wrap" => matches!(
value.to_ascii_lowercase().as_str(),
"normal" | "break-word" | "anywhere"
),
"word-break" => matches!(
value.to_ascii_lowercase().as_str(),
"normal" | "break-all" | "keep-all" | "break-word"
),
"text-wrap" => matches!( "text-wrap" => matches!(
value.to_ascii_lowercase().as_str(), value.to_ascii_lowercase().as_str(),
"auto" | "wrap" | "balance" | "wrap balance" | "balance wrap" "auto" | "wrap" | "balance" | "wrap balance" | "balance wrap"
@@ -6460,6 +6493,55 @@ mod tests {
assert!(!supports_declaration("text-indent", "10px hanging")); assert!(!supports_declaration("text-indent", "10px hanging"));
} }
#[test]
fn text_break_longhands_parse_alias_and_css_wide_values() {
let values = compute_style(
"div",
Some(
"overflow-wrap:break-word;\
word-wrap:anywhere;\
word-break:break-all",
),
);
assert_eq!(values.overflow_wrap, Some(crate::OverflowWrap::Anywhere));
assert_eq!(values.word_break, Some(crate::WordBreak::BreakAll));
let legacy = compute_style("div", Some("word-break:break-word"));
assert_eq!(legacy.word_break, Some(crate::WordBreak::BreakWord));
for keyword in ["inherit", "unset", "revert", "revert-layer"] {
let style = compute_style(
"div",
Some(&format!(
"overflow-wrap:anywhere;overflow-wrap:{keyword};\
word-break:break-all;word-break:{keyword}"
)),
);
assert_eq!(style.overflow_wrap, None, "{keyword}");
assert_eq!(style.word_break, None, "{keyword}");
}
let initial = compute_style(
"div",
Some(
"overflow-wrap:anywhere;overflow-wrap:initial;\
word-break:break-all;word-break:initial",
),
);
assert_eq!(initial.overflow_wrap, Some(crate::OverflowWrap::Normal));
assert_eq!(initial.word_break, Some(crate::WordBreak::Normal));
for property in ["overflow-wrap", "word-wrap"] {
assert!(supports_declaration(property, "normal"));
assert!(supports_declaration(property, "break-word"));
assert!(supports_declaration(property, "anywhere"));
assert!(!supports_declaration(property, "break-all"));
}
for value in ["normal", "break-all", "keep-all", "break-word"] {
assert!(supports_declaration("word-break", value), "{value}");
}
assert!(!supports_declaration("word-break", "anywhere"));
}
#[test] #[test]
fn nonzero_unitless_font_size_is_invalid() { fn nonzero_unitless_font_size_is_invalid() {
let inherited = compute_style("div", Some("font-size:14px;font-size:.813")); let inherited = compute_style("div", Some("font-size:14px;font-size:.813"));
+57
View File
@@ -11,6 +11,51 @@ use crate::{CacheKeyFlags, Metrics};
pub use fontdb::{Family, Stretch, Style, Weight}; pub use fontdb::{Family, Stretch, Style, Weight};
/// CSS-compatible line-breaking policy attached to a rich-text span.
///
/// This is optional on [`Attrs`] so existing cosmic-text users retain the
/// library's ordinary `Wrap` behavior. Browser integrations can opt in when
/// line-breaking rules vary inside one shaped paragraph.
#[derive(Clone, Copy, Debug, Eq, Hash, PartialEq)]
pub struct CssLineBreak {
/// Whether soft wrapping is enabled for this span.
pub wrap: bool,
/// CSS `word-break` behavior.
pub word_break: CssWordBreak,
/// CSS `overflow-wrap` behavior.
pub overflow_wrap: CssOverflowWrap,
}
impl Default for CssLineBreak {
fn default() -> Self {
Self {
wrap: true,
word_break: CssWordBreak::Normal,
overflow_wrap: CssOverflowWrap::Normal,
}
}
}
/// CSS `word-break` values used by the line-breaking adapter.
#[derive(Clone, Copy, Debug, Default, Eq, Hash, PartialEq)]
pub enum CssWordBreak {
#[default]
Normal,
BreakAll,
KeepAll,
/// Legacy `word-break: break-word` compatibility behavior.
BreakWord,
}
/// CSS `overflow-wrap` values used by the line-breaking adapter.
#[derive(Clone, Copy, Debug, Default, Eq, Hash, PartialEq)]
pub enum CssOverflowWrap {
#[default]
Normal,
BreakWord,
Anywhere,
}
/// Text color /// Text color
#[derive(Clone, Copy, Debug, PartialOrd, Ord, Eq, Hash, PartialEq)] #[derive(Clone, Copy, Debug, PartialOrd, Ord, Eq, Hash, PartialEq)]
pub struct Color(pub u32); pub struct Color(pub u32);
@@ -349,6 +394,8 @@ pub struct Attrs<'a> {
pub letter_spacing_opt: Option<LetterSpacing>, pub letter_spacing_opt: Option<LetterSpacing>,
pub font_features: FontFeatures, pub font_features: FontFeatures,
pub font_variations: FontVariations, pub font_variations: FontVariations,
/// Optional browser-grade line-breaking policy for this text span.
pub css_line_break: Option<CssLineBreak>,
} }
impl<'a> Attrs<'a> { impl<'a> Attrs<'a> {
@@ -372,6 +419,7 @@ impl<'a> Attrs<'a> {
letter_spacing_opt: None, letter_spacing_opt: None,
font_features: FontFeatures::new(), font_features: FontFeatures::new(),
font_variations: FontVariations::new(), font_variations: FontVariations::new(),
css_line_break: None,
} }
} }
@@ -468,6 +516,12 @@ impl<'a> Attrs<'a> {
self self
} }
/// Attach CSS line-breaking behavior to this span.
pub fn css_line_break(mut self, policy: CssLineBreak) -> Self {
self.css_line_break = Some(policy);
self
}
/// Check if font matches /// Check if font matches
pub fn matches(&self, face: &fontdb::FaceInfo) -> bool { pub fn matches(&self, face: &fontdb::FaceInfo) -> bool {
//TODO: smarter way of including emoji //TODO: smarter way of including emoji
@@ -529,6 +583,7 @@ pub struct AttrsOwned {
pub letter_spacing_opt: Option<LetterSpacing>, pub letter_spacing_opt: Option<LetterSpacing>,
pub font_features: FontFeatures, pub font_features: FontFeatures,
pub font_variations: FontVariations, pub font_variations: FontVariations,
pub css_line_break: Option<CssLineBreak>,
} }
impl AttrsOwned { impl AttrsOwned {
@@ -549,6 +604,7 @@ impl AttrsOwned {
letter_spacing_opt: attrs.letter_spacing_opt, letter_spacing_opt: attrs.letter_spacing_opt,
font_features: attrs.font_features.clone(), font_features: attrs.font_features.clone(),
font_variations: attrs.font_variations.clone(), font_variations: attrs.font_variations.clone(),
css_line_break: attrs.css_line_break,
} }
} }
@@ -569,6 +625,7 @@ impl AttrsOwned {
letter_spacing_opt: self.letter_spacing_opt, letter_spacing_opt: self.letter_spacing_opt,
font_features: self.font_features.clone(), font_features: self.font_features.clone(),
font_variations: self.font_variations.clone(), font_variations: self.font_variations.clone(),
css_line_break: self.css_line_break,
} }
} }
} }
+5
View File
@@ -120,6 +120,10 @@ pub enum Wrap {
Word, Word,
/// Wraps at the word level, or fallback to glyph level if a word can't fit on a line by itself /// Wraps at the word level, or fallback to glyph level if a word can't fit on a line by itself
WordOrGlyph, WordOrGlyph,
/// Browser min-content wrapping. Like [`Self::WordOrGlyph`], but only
/// span boundaries contributed by `overflow-wrap:anywhere` (and legacy
/// `word-break:break-word`) participate.
WordOrGlyphMinContent,
} }
impl Display for Wrap { impl Display for Wrap {
@@ -128,6 +132,7 @@ impl Display for Wrap {
Self::None => write!(f, "No Wrap"), Self::None => write!(f, "No Wrap"),
Self::Word => write!(f, "Word Wrap"), Self::Word => write!(f, "Word Wrap"),
Self::WordOrGlyph => write!(f, "Word Wrap or Character"), Self::WordOrGlyph => write!(f, "Word Wrap or Character"),
Self::WordOrGlyphMinContent => write!(f, "Min-content Word Wrap or Character"),
Self::Glyph => write!(f, "Character"), Self::Glyph => write!(f, "Character"),
} }
} }
+429 -28
View File
@@ -8,13 +8,14 @@ use core::cmp::{max, min};
use core::fmt; use core::fmt;
use core::mem; use core::mem;
use core::ops::Range; use core::ops::Range;
use unicode_linebreak::{break_property, linebreaks, BreakClass};
use unicode_script::{Script, UnicodeScript}; use unicode_script::{Script, UnicodeScript};
use unicode_segmentation::UnicodeSegmentation; use unicode_segmentation::UnicodeSegmentation;
use crate::fallback::FontFallbackIter; use crate::fallback::FontFallbackIter;
use crate::{ use crate::{
math, Align, AttrsList, CacheKeyFlags, Color, Font, FontSystem, LayoutGlyph, LayoutLine, math, Align, AttrsList, CacheKeyFlags, Color, CssLineBreak, CssOverflowWrap, CssWordBreak,
Metrics, Wrap, Font, FontSystem, LayoutGlyph, LayoutLine, Metrics, Wrap,
}; };
/// The shaping strategy of some text. /// The shaping strategy of some text.
@@ -687,6 +688,16 @@ impl ShapeGlyph {
pub struct ShapeWord { pub struct ShapeWord {
pub blank: bool, pub blank: bool,
pub glyphs: Vec<ShapeGlyph>, pub glyphs: Vec<ShapeGlyph>,
/// Additional ordinary soft-wrap positions inside this UAX#14 word.
/// Values are glyph indices and therefore never split a shaped cluster.
soft_breaks: Vec<usize>,
/// Emergency positions used by constrained layout.
emergency_breaks: Vec<usize>,
/// Emergency positions that also contribute to min-content.
min_content_breaks: Vec<usize>,
/// Whether this word came from CSS-policy-bearing rich text. When false,
/// cosmic-text's historical `WordOrGlyph` behavior remains unchanged.
custom_line_breaks: bool,
} }
impl ShapeWord { impl ShapeWord {
@@ -697,9 +708,92 @@ impl ShapeWord {
Self { Self {
blank: true, blank: true,
glyphs: Vec::default(), glyphs: Vec::default(),
soft_breaks: Vec::new(),
emergency_breaks: Vec::new(),
min_content_breaks: Vec::new(),
custom_line_breaks: false,
} }
} }
fn glyph_indices_for_offsets(&self, mut offsets: Vec<usize>) -> Vec<usize> {
offsets.sort_unstable();
offsets.dedup();
(1..self.glyphs.len())
.filter(|&index| {
let before = &self.glyphs[index - 1];
let after = &self.glyphs[index];
let offset = if before.end <= after.start {
after.start
} else if after.end <= before.start {
before.start
} else {
return false;
};
offsets.binary_search(&offset).is_ok()
})
.collect()
}
fn set_line_breaks(
&mut self,
custom: bool,
soft_offsets: impl Iterator<Item = usize>,
emergency_offsets: impl Iterator<Item = usize>,
min_content_offsets: impl Iterator<Item = usize>,
) {
let soft_breaks = self.glyph_indices_for_offsets(soft_offsets.collect());
let emergency_breaks = self.glyph_indices_for_offsets(emergency_offsets.collect());
let min_content_breaks =
self.glyph_indices_for_offsets(min_content_offsets.collect());
self.custom_line_breaks = custom;
self.soft_breaks = soft_breaks;
self.emergency_breaks = emergency_breaks;
self.min_content_breaks = min_content_breaks;
}
fn reverse_glyphs(&mut self) {
self.glyphs.reverse();
let len = self.glyphs.len();
for breaks in [
&mut self.soft_breaks,
&mut self.emergency_breaks,
&mut self.min_content_breaks,
] {
for index in breaks.iter_mut() {
*index = len.saturating_sub(*index);
}
breaks.sort_unstable();
}
}
fn cluster_boundaries(&self) -> Vec<usize> {
(1..self.glyphs.len())
.filter(|&index| {
let before = &self.glyphs[index - 1];
let after = &self.glyphs[index];
before.end <= after.start || after.end <= before.start
})
.collect()
}
fn break_indices(&self, wrap: Wrap, use_emergency: bool) -> Vec<usize> {
if wrap == Wrap::Glyph || (!self.custom_line_breaks && use_emergency) {
return self.cluster_boundaries();
}
let mut breaks = self.soft_breaks.clone();
if use_emergency {
let emergency = if wrap == Wrap::WordOrGlyphMinContent {
&self.min_content_breaks
} else {
&self.emergency_breaks
};
breaks.extend(emergency.iter().copied());
breaks.sort_unstable();
breaks.dedup();
}
breaks
}
/// Shape a word into a set of glyphs. /// Shape a word into a set of glyphs.
#[allow(clippy::too_many_arguments)] #[allow(clippy::too_many_arguments)]
pub fn new( pub fn new(
@@ -785,6 +879,10 @@ impl ShapeWord {
self.blank = blank; self.blank = blank;
self.glyphs = glyphs; self.glyphs = glyphs;
self.soft_breaks.clear();
self.emergency_breaks.clear();
self.min_content_breaks.clear();
self.custom_line_breaks = false;
} }
/// Get the width of the [`ShapeWord`] in pixels, using the [`ShapeGlyph::width`] function. /// Get the width of the [`ShapeWord`] in pixels, using the [`ShapeGlyph::width`] function.
@@ -797,6 +895,219 @@ impl ShapeWord {
} }
} }
#[derive(Clone, Copy)]
struct GraphemeBreakData {
end: usize,
/// The unmodified UAX#14 class, used by `keep-all`.
class: BreakClass,
/// Blink's break-all iterator tailors a small number of characters before
/// consulting its pair table. Keep that tailoring out of `keep-all`.
break_all_class: BreakClass,
vertical_line: bool,
policy: Option<CssLineBreak>,
}
fn grapheme_break_class(grapheme: &str) -> BreakClass {
grapheme
.chars()
.map(|ch| break_property(ch as u32))
.find(|class| {
!matches!(
class,
BreakClass::CombiningMark | BreakClass::ZeroWidthJoiner
)
})
.unwrap_or(BreakClass::CombiningMark)
}
/// Letter/number-like UAX#14 classes governed by `word-break`'s "within
/// words" rules. Complex-context scripts deliberately stay out of keep-all,
/// matching Blink's exemption for Southeast Asian dictionary breaking.
fn keep_all_word_class(class: BreakClass) -> bool {
matches!(
class,
BreakClass::Alphabetic
| BreakClass::Ambiguous
| BreakClass::HebrewLetter
| BreakClass::Numeric
| BreakClass::Ideographic
| BreakClass::ConditionalJapaneseStarter
| BreakClass::HangulLvSyllable
| BreakClass::HangulLvtSyllable
| BreakClass::HangulLJamo
| BreakClass::HangulVJamo
| BreakClass::HangulTJamo
)
}
/// CSS `break-all` adds opportunities between typographic letter units. The
/// surrounding UAX#14 iterator continues to own punctuation, emoji, spaces,
/// joiners, and mandatory breaks; this is the key distinction from terminal
/// style "break character" wrapping.
fn break_all_class(grapheme: &str) -> BreakClass {
// Blink tailors PLUS SIGN to the alphabetic class for break-all.
if grapheme == "+" {
BreakClass::Alphabetic
} else {
grapheme_break_class(grapheme)
}
}
fn break_all_pair(before: GraphemeBreakData, after: GraphemeBreakData) -> bool {
use BreakClass::{
After, Alphabetic, Ambiguous, CloseParenthesis, ClosePunctuation, ComplexContext,
Exclamation, HebrewLetter, Hyphen, InfixSeparator, Numeric, OpenPunctuation, Postfix,
Prefix, Symbol,
};
let before_class = before.break_all_class;
let after_class = after.break_all_class;
let after_vertical_line = after_class == After && after.vertical_line;
let al_like_after = matches!(
after_class,
Ambiguous
| Alphabetic
| Hyphen
| Numeric
| OpenPunctuation
| Prefix
| ComplexContext
| HebrewLetter
) || after_vertical_line;
match before_class {
Ambiguous | Alphabetic | After | Numeric | ComplexContext | Symbol | HebrewLetter => {
al_like_after
}
ClosePunctuation => {
matches!(
after_class,
Ambiguous | Alphabetic | Hyphen | Numeric | Prefix | HebrewLetter
) || after_vertical_line
}
CloseParenthesis => {
matches!(
after_class,
Ambiguous
| Alphabetic
| Hyphen
| Numeric
| Prefix
| ComplexContext
| HebrewLetter
) || after_vertical_line
}
Exclamation | Postfix => matches!(
after_class,
Ambiguous
| Alphabetic
| Hyphen
| Numeric
| Postfix
| Prefix
| HebrewLetter
) || after_vertical_line,
InfixSeparator => {
matches!(
after_class,
Ambiguous | Alphabetic | Hyphen | Numeric | HebrewLetter
) || after_vertical_line
}
Hyphen => after_class == Numeric,
Prefix => after_class == Postfix,
_ => false,
}
}
fn blink_normal_break(
normal_break: bool,
before: GraphemeBreakData,
after: GraphemeBreakData,
) -> bool {
if !normal_break {
return false;
}
// unicode-linebreak follows the default UAX#14 pair table, while Blink's
// ICU iterator keeps ASCII solidus and vertical line inside ordinary word
// runs (`a/b`, `a|b`). Blink's break-all tailoring then adds the narrower
// opportunities explicitly. Keep this browser-specific adjustment behind
// CssLineBreak so cosmic-text's existing non-browser behavior is stable.
!(before.vertical_line
|| (before.class == BreakClass::Symbol && keep_all_word_class(after.class)))
}
struct CssBreakData {
word_breaks: Vec<usize>,
soft_breaks: Vec<usize>,
emergency_breaks: Vec<usize>,
min_content_breaks: Vec<usize>,
}
fn css_break_data(span: &str, span_start: usize, attrs_list: &AttrsList) -> CssBreakData {
let normal_breaks: Vec<usize> = linebreaks(span).map(|(offset, _)| offset).collect();
let graphemes: Vec<GraphemeBreakData> = span
.grapheme_indices(true)
.map(|(start, grapheme)| {
let class = grapheme_break_class(grapheme);
GraphemeBreakData {
end: start + grapheme.len(),
class,
break_all_class: break_all_class(grapheme),
vertical_line: grapheme == "|",
policy: attrs_list.get_span(span_start + start).css_line_break,
}
})
.collect();
let mut data = CssBreakData {
word_breaks: Vec::new(),
soft_breaks: Vec::new(),
emergency_breaks: Vec::new(),
min_content_breaks: Vec::new(),
};
for pair in graphemes.windows(2) {
let before = pair[0];
let after = pair[1];
let offset = before.end;
let absolute = span_start + offset;
let normal_break = normal_breaks.binary_search(&offset).is_ok();
let Some(policy) = before.policy else {
if normal_break {
data.word_breaks.push(offset);
}
continue;
};
if !policy.wrap {
continue;
}
let keep_all = policy.word_break == CssWordBreak::KeepAll
&& keep_all_word_class(before.class)
&& keep_all_word_class(after.class);
if blink_normal_break(normal_break, before, after) && !keep_all {
data.word_breaks.push(offset);
} else if policy.word_break == CssWordBreak::BreakAll && break_all_pair(before, after) {
data.soft_breaks.push(absolute);
}
let anywhere = policy.word_break == CssWordBreak::BreakWord
|| policy.overflow_wrap == CssOverflowWrap::Anywhere;
let emergency = anywhere || policy.overflow_wrap == CssOverflowWrap::BreakWord;
if emergency {
data.emergency_breaks.push(absolute);
}
if anywhere {
data.min_content_breaks.push(absolute);
}
}
// ShapeLine represents one mandatory-line segment. Its end must terminate
// the final word even when the preceding CSS span disables soft wrapping.
data.word_breaks.push(span.len());
data.word_breaks.sort_unstable();
data.word_breaks.dedup();
data
}
/// A shaped span (for bidirectional processing) /// A shaped span (for bidirectional processing)
#[derive(Clone, Debug)] #[derive(Clone, Debug)]
pub struct ShapeSpan { pub struct ShapeSpan {
@@ -871,8 +1182,9 @@ impl ShapeSpan {
cached_words.extend(words.drain(..).rev()); cached_words.extend(words.drain(..).rev());
} }
let breaks = css_break_data(span, span_range.start, attrs_list);
let mut start_word = 0; let mut start_word = 0;
for (end_lb, _) in unicode_linebreak::linebreaks(span) { for end_lb in breaks.word_breaks.iter().copied() {
let mut start_lb = end_lb; let mut start_lb = end_lb;
for (i, c) in span[start_word..end_lb].char_indices().rev() { for (i, c) in span[start_word..end_lb].char_indices().rev() {
// TODO: Not all whitespace characters are linebreakable, e.g. 00A0 (No-break // TODO: Not all whitespace characters are linebreakable, e.g. 00A0 (No-break
@@ -896,6 +1208,43 @@ impl ShapeSpan {
false, false,
shaping, shaping,
); );
let absolute = (span_range.start + start_word)..(span_range.start + start_lb);
let custom = line[absolute.clone()]
.char_indices()
.any(|(offset, _)| {
attrs_list
.get_span(absolute.start + offset)
.css_line_break
.is_some()
});
let soft_start = breaks
.soft_breaks
.partition_point(|offset| *offset <= absolute.start);
let soft_end = breaks
.soft_breaks
.partition_point(|offset| *offset < absolute.end);
let emergency_start = breaks
.emergency_breaks
.partition_point(|offset| *offset <= absolute.start);
let emergency_end = breaks
.emergency_breaks
.partition_point(|offset| *offset < absolute.end);
let min_start = breaks
.min_content_breaks
.partition_point(|offset| *offset <= absolute.start);
let min_end = breaks
.min_content_breaks
.partition_point(|offset| *offset < absolute.end);
word.set_line_breaks(
custom,
breaks.soft_breaks[soft_start..soft_end].iter().copied(),
breaks.emergency_breaks[emergency_start..emergency_end]
.iter()
.copied(),
breaks.min_content_breaks[min_start..min_end]
.iter()
.copied(),
);
words.push(word); words.push(word);
} }
if start_lb < end_lb { if start_lb < end_lb {
@@ -912,6 +1261,18 @@ impl ShapeSpan {
true, true,
shaping, shaping,
); );
let absolute = (span_range.start + start_lb + i)
..(span_range.start + start_lb + i + c.len_utf8());
let custom = attrs_list
.get_span(absolute.start)
.css_line_break
.is_some();
word.set_line_breaks(
custom,
core::iter::empty(),
core::iter::empty(),
core::iter::empty(),
);
words.push(word); words.push(word);
} }
} }
@@ -921,7 +1282,7 @@ impl ShapeSpan {
// Reverse glyphs in RTL lines // Reverse glyphs in RTL lines
if line_rtl { if line_rtl {
for word in &mut words { for word in &mut words {
word.glyphs.reverse(); word.reverse_glyphs();
} }
} }
@@ -1338,13 +1699,21 @@ impl ShapeLine {
} }
word_range_width += word_width; word_range_width += word_width;
continue; continue;
} else if wrap == Wrap::Glyph } else {
// Make sure that the word is able to fit on it's own line, if not, fall back to Glyph wrapping. let emergency = wrap == Wrap::Glyph
|| (wrap == Wrap::WordOrGlyph && word_width > width_opt.unwrap_or(f32::INFINITY)) || (matches!(
{ wrap,
Wrap::WordOrGlyph | Wrap::WordOrGlyphMinContent
) && word_width > width_opt.unwrap_or(f32::INFINITY));
let break_indices = word.break_indices(wrap, emergency);
if !break_indices.is_empty() {
// Commit the current line so that the word starts on the next line. // Commit the current line so that the word starts on the next line.
if word_range_width > 0. if word_range_width > 0.
&& wrap == Wrap::WordOrGlyph && word.soft_breaks.is_empty()
&& matches!(
wrap,
Wrap::WordOrGlyph | Wrap::WordOrGlyphMinContent
)
&& word_width > width_opt.unwrap_or(f32::INFINITY) && word_width > width_opt.unwrap_or(f32::INFINITY)
{ {
add_to_visual_line( add_to_visual_line(
@@ -1365,18 +1734,29 @@ impl ShapeLine {
fitting_start = (i, 0); fitting_start = (i, 0);
} }
for (glyph_i, glyph) in word.glyphs.iter().enumerate().rev() { let mut boundaries = Vec::with_capacity(break_indices.len() + 2);
let glyph_width = glyph.width(font_size); boundaries.push(0);
if current_visual_line.w + (word_range_width + glyph_width) boundaries.extend(break_indices);
boundaries.push(word.glyphs.len());
for chunk in boundaries.windows(2).rev() {
let start = chunk[0];
let end = chunk[1];
let chunk_width = word.glyphs[start..end]
.iter()
.map(|glyph| glyph.width(font_size))
.sum::<f32>();
if current_visual_line.w + (word_range_width + chunk_width)
<= width_opt.unwrap_or(f32::INFINITY) <= width_opt.unwrap_or(f32::INFINITY)
|| (current_visual_line.ranges.is_empty()
&& word_range_width == 0.)
{ {
word_range_width += glyph_width; word_range_width += chunk_width;
continue; continue;
} else { } else {
add_to_visual_line( add_to_visual_line(
&mut current_visual_line, &mut current_visual_line,
span_index, span_index,
(i, glyph_i + 1), (i, end),
fitting_start, fitting_start,
word_range_width, word_range_width,
number_of_blanks, number_of_blanks,
@@ -1386,8 +1766,8 @@ impl ShapeLine {
cached_visual_lines.pop().unwrap_or_default(); cached_visual_lines.pop().unwrap_or_default();
number_of_blanks = 0; number_of_blanks = 0;
word_range_width = glyph_width; word_range_width = chunk_width;
fitting_start = (i, glyph_i + 1); fitting_start = (i, end);
} }
} }
} else { } else {
@@ -1437,6 +1817,7 @@ impl ShapeLine {
} }
} }
} }
}
add_to_visual_line( add_to_visual_line(
&mut current_visual_line, &mut current_visual_line,
span_index, span_index,
@@ -1464,13 +1845,21 @@ impl ShapeLine {
} }
word_range_width += word_width; word_range_width += word_width;
continue; continue;
} else if wrap == Wrap::Glyph } else {
// Make sure that the word is able to fit on it's own line, if not, fall back to Glyph wrapping. let emergency = wrap == Wrap::Glyph
|| (wrap == Wrap::WordOrGlyph && word_width > width_opt.unwrap_or(f32::INFINITY)) || (matches!(
{ wrap,
Wrap::WordOrGlyph | Wrap::WordOrGlyphMinContent
) && word_width > width_opt.unwrap_or(f32::INFINITY));
let break_indices = word.break_indices(wrap, emergency);
if !break_indices.is_empty() {
// Commit the current line so that the word starts on the next line. // Commit the current line so that the word starts on the next line.
if word_range_width > 0. if word_range_width > 0.
&& wrap == Wrap::WordOrGlyph && word.soft_breaks.is_empty()
&& matches!(
wrap,
Wrap::WordOrGlyph | Wrap::WordOrGlyphMinContent
)
&& word_width > width_opt.unwrap_or(f32::INFINITY) && word_width > width_opt.unwrap_or(f32::INFINITY)
{ {
add_to_visual_line( add_to_visual_line(
@@ -1491,19 +1880,30 @@ impl ShapeLine {
fitting_start = (i, 0); fitting_start = (i, 0);
} }
for (glyph_i, glyph) in word.glyphs.iter().enumerate() { let mut boundaries = Vec::with_capacity(break_indices.len() + 2);
let glyph_width = glyph.width(font_size); boundaries.push(0);
if current_visual_line.w + (word_range_width + glyph_width) boundaries.extend(break_indices);
boundaries.push(word.glyphs.len());
for chunk in boundaries.windows(2) {
let start = chunk[0];
let end = chunk[1];
let chunk_width = word.glyphs[start..end]
.iter()
.map(|glyph| glyph.width(font_size))
.sum::<f32>();
if current_visual_line.w + (word_range_width + chunk_width)
<= width_opt.unwrap_or(f32::INFINITY) <= width_opt.unwrap_or(f32::INFINITY)
|| (current_visual_line.ranges.is_empty()
&& word_range_width == 0.)
{ {
word_range_width += glyph_width; word_range_width += chunk_width;
continue; continue;
} else { } else {
add_to_visual_line( add_to_visual_line(
&mut current_visual_line, &mut current_visual_line,
span_index, span_index,
fitting_start, fitting_start,
(i, glyph_i), (i, start),
word_range_width, word_range_width,
number_of_blanks, number_of_blanks,
); );
@@ -1512,8 +1912,8 @@ impl ShapeLine {
cached_visual_lines.pop().unwrap_or_default(); cached_visual_lines.pop().unwrap_or_default();
number_of_blanks = 0; number_of_blanks = 0;
word_range_width = glyph_width; word_range_width = chunk_width;
fitting_start = (i, glyph_i); fitting_start = (i, start);
} }
} }
} else { } else {
@@ -1560,6 +1960,7 @@ impl ShapeLine {
} }
} }
} }
}
add_to_visual_line( add_to_visual_line(
&mut current_visual_line, &mut current_visual_line,
span_index, span_index,