Jump to source and preview

3 files changed, 91 insertions, 14 deletions

diff --git a/library/src/layout/par.rs b/library/src/layout/par.rs
index 8dd81d29..6178c059 100644
--- a/library/src/layout/par.rs
+++ b/library/src/layout/par.rs
@@ -145,12 +145,12 @@ impl ParNode {
             let children = par.children();
 
             // Collect all text into one string for BiDi analysis.
-            let (text, segments) = collect(&children, &styles, consecutive)?;
+            let (text, segments, spans) = collect(&children, &styles, consecutive)?;
 
             // Perform BiDi analysis and then prepare paragraph layout by building a
             // representation on which we can do line breaking without layouting
             // each and every line from scratch.
-            let p = prepare(&mut vt, &children, &text, segments, styles, region)?;
+            let p = prepare(&mut vt, &children, &text, segments, spans, styles, region)?;
 
             // Break the paragraph into lines.
             let lines = linebreak(&vt, &p, region.x);
@@ -264,6 +264,8 @@ struct Preparation<'a> {
     bidi: BidiInfo<'a>,
     /// Text runs, spacing and layouted nodes.
     items: Vec<Item<'a>>,
+    /// The span mapper.
+    spans: SpanMapper,
     /// The styles shared by all children.
     styles: StyleChain<'a>,
     /// Whether to hyphenate if it's the same for all children.
@@ -388,6 +390,35 @@ impl<'a> Item<'a> {
     }
 }
 
+/// Maps byte offsets back to spans.
+pub struct SpanMapper(Vec<(usize, Span)>);
+
+impl SpanMapper {
+    /// Create a new span mapper.
+    pub fn new() -> Self {
+        Self(vec![])
+    }
+
+    /// Push a span for a segment with the given length.
+    pub fn push(&mut self, len: usize, span: Span) {
+        self.0.push((len, span));
+    }
+
+    /// Determine the span at the given byte offset.
+    ///
+    /// May return a detached span.
+    pub fn span_at(&self, offset: usize) -> (Span, u16) {
+        let mut cursor = 0;
+        for &(len, span) in &self.0 {
+            if (cursor..=cursor + len).contains(&offset) {
+                return (span, u16::try_from(offset - cursor).unwrap_or(0));
+            }
+            cursor += len;
+        }
+        (Span::detached(), 0)
+    }
+}
+
 /// A layouted line, consisting of a sequence of layouted paragraph items that
 /// are mostly borrowed from the preparation phase. This type enables you to
 /// measure the size of a line in a range before comitting to building the
@@ -485,10 +516,11 @@ fn collect<'a>(
     children: &'a [Content],
     styles: &'a StyleChain<'a>,
     consecutive: bool,
-) -> SourceResult<(String, Vec<(Segment<'a>, StyleChain<'a>)>)> {
+) -> SourceResult<(String, Vec<(Segment<'a>, StyleChain<'a>)>, SpanMapper)> {
     let mut full = String::new();
     let mut quoter = Quoter::new();
     let mut segments = vec![];
+    let mut spans = SpanMapper::new();
     let mut iter = children.iter().peekable();
 
     if consecutive {
@@ -578,6 +610,8 @@ fn collect<'a>(
             quoter.last(last);
         }
 
+        spans.push(segment.len(), child.span());
+
         if let (Some((Segment::Text(last_len), last_styles)), Segment::Text(len)) =
             (segments.last_mut(), segment)
         {
@@ -590,7 +624,7 @@ fn collect<'a>(
         segments.push((segment, styles));
     }
 
-    Ok((full, segments))
+    Ok((full, segments, spans))
 }
 
 /// Prepare paragraph layout by shaping the whole paragraph and layouting all
@@ -600,6 +634,7 @@ fn prepare<'a>(
     children: &'a [Content],
     text: &'a str,
     segments: Vec<(Segment<'a>, StyleChain<'a>)>,
+    spans: SpanMapper,
     styles: StyleChain<'a>,
     region: Size,
 ) -> SourceResult<Preparation<'a>> {
@@ -620,7 +655,7 @@ fn prepare<'a>(
         let end = cursor + segment.len();
         match segment {
             Segment::Text(_) => {
-                shape_range(&mut items, vt, &bidi, cursor..end, styles);
+                shape_range(&mut items, vt, &bidi, cursor..end, &spans, styles);
             }
             Segment::Spacing(spacing) => match spacing {
                 Spacing::Rel(v) => {
@@ -655,6 +690,7 @@ fn prepare<'a>(
     Ok(Preparation {
         bidi,
         items,
+        spans,
         styles,
         hyphenate: shared_get(styles, children, TextNode::hyphenate_in),
         lang: shared_get(styles, children, TextNode::lang_in),
@@ -670,11 +706,12 @@ fn shape_range<'a>(
     vt: &Vt,
     bidi: &BidiInfo<'a>,
     range: Range,
+    spans: &SpanMapper,
     styles: StyleChain<'a>,
 ) {
-    let mut process = |text, level: BidiLevel| {
+    let mut process = |range: Range, level: BidiLevel| {
         let dir = if level.is_ltr() { Dir::LTR } else { Dir::RTL };
-        let shaped = shape(vt, text, styles, dir);
+        let shaped = shape(vt, range.start, &bidi.text[range], spans, styles, dir);
         items.push(Item::Text(shaped));
     };
 
@@ -694,7 +731,7 @@ fn shape_range<'a>(
 
         if level != prev_level || !is_compatible(script, prev_script) {
             if cursor < i {
-                process(&bidi.text[cursor..i], prev_level);
+                process(cursor..i, prev_level);
             }
             cursor = i;
             prev_level = level;
@@ -704,7 +741,7 @@ fn shape_range<'a>(
         }
     }
 
-    process(&bidi.text[cursor..range.end], prev_level);
+    process(cursor..range.end, prev_level);
 }
 
 /// Whether this is not a specific script.
@@ -1073,7 +1110,7 @@ fn line<'a>(
         if hyphen || start + shaped.text.len() > range.end {
             if hyphen || start < range.end || before.is_empty() {
                 let shifted = start - base..range.end - base;
-                let mut reshaped = shaped.reshape(vt, shifted);
+                let mut reshaped = shaped.reshape(vt, &p.spans, shifted);
                 if hyphen || shy {
                     reshaped.push_hyphen(vt);
                 }
@@ -1096,7 +1133,7 @@ fn line<'a>(
         if range.start + shaped.text.len() > end {
             if range.start < end {
                 let shifted = range.start - base..end - base;
-                let reshaped = shaped.reshape(vt, shifted);
+                let reshaped = shaped.reshape(vt, &p.spans, shifted);
                 width += reshaped.width;
                 first = Some(Item::Text(reshaped));
             }
diff --git a/library/src/math/fragment.rs b/library/src/math/fragment.rs
index 73daa4b2..de456b82 100644
--- a/library/src/math/fragment.rs
+++ b/library/src/math/fragment.rs
@@ -222,6 +222,8 @@ impl GlyphFragment {
                 c: self.c,
                 x_advance: Em::from_length(self.width, self.font_size),
                 x_offset: Em::zero(),
+                span: Span::detached(),
+                offset: 0,
             }],
         };
         let size = Size::new(self.width, self.ascent + self.descent);
diff --git a/library/src/text/shaping.rs b/library/src/text/shaping.rs
index e7ce4027..a96238d9 100644
--- a/library/src/text/shaping.rs
+++ b/library/src/text/shaping.rs
@@ -6,6 +6,7 @@ use typst::font::{Font, FontVariant};
 use typst::util::SliceExt;
 
 use super::*;
+use crate::layout::SpanMapper;
 use crate::prelude::*;
 
 /// The result of shaping text.
@@ -14,6 +15,8 @@ use crate::prelude::*;
 /// measured, used to reshape substrings more quickly and converted into a
 /// frame.
 pub struct ShapedText<'a> {
+    /// The start of the text in the full paragraph.
+    pub base: usize,
     /// The text that was shaped.
     pub text: &'a str,
     /// The text direction.
@@ -53,6 +56,10 @@ pub struct ShapedGlyph {
     pub safe_to_break: bool,
     /// The first char in this glyph's cluster.
     pub c: char,
+    /// The source code location of the text.
+    pub span: Span,
+    /// The offset within the spanned text.
+    pub offset: u16,
 }
 
 impl ShapedGlyph {
@@ -110,6 +117,8 @@ impl<'a> ShapedText<'a> {
                         },
                     x_offset: glyph.x_offset,
                     c: glyph.c,
+                    span: glyph.span,
+                    offset: glyph.offset,
                 })
                 .collect();
 
@@ -187,9 +196,15 @@ impl<'a> ShapedText<'a> {
 
     /// Reshape a range of the shaped text, reusing information from this
     /// shaping process if possible.
-    pub fn reshape(&'a self, vt: &Vt, text_range: Range<usize>) -> ShapedText<'a> {
+    pub fn reshape(
+        &'a self,
+        vt: &Vt,
+        spans: &SpanMapper,
+        text_range: Range<usize>,
+    ) -> ShapedText<'a> {
         if let Some(glyphs) = self.slice_safe_to_break(text_range.clone()) {
             Self {
+                base: self.base + text_range.start,
                 text: &self.text[text_range],
                 dir: self.dir,
                 styles: self.styles,
@@ -199,7 +214,14 @@ impl<'a> ShapedText<'a> {
                 glyphs: Cow::Borrowed(glyphs),
             }
         } else {
-            shape(vt, &self.text[text_range], self.styles, self.dir)
+            shape(
+                vt,
+                self.base + text_range.start,
+                &self.text[text_range],
+                spans,
+                self.styles,
+                self.dir,
+            )
         }
     }
 
@@ -225,6 +247,8 @@ impl<'a> ShapedText<'a> {
                 cluster,
                 safe_to_break: true,
                 c: '-',
+                span: Span::detached(),
+                offset: 0,
             });
             Some(())
         });
@@ -298,6 +322,8 @@ impl Debug for ShapedText<'_> {
 /// Holds shaping results and metadata common to all shaped segments.
 struct ShapingContext<'a> {
     vt: &'a Vt<'a>,
+    base: usize,
+    spans: &'a SpanMapper,
     glyphs: Vec<ShapedGlyph>,
     used: Vec<Font>,
     styles: StyleChain<'a>,
@@ -311,13 +337,17 @@ struct ShapingContext<'a> {
 /// Shape text into [`ShapedText`].
 pub fn shape<'a>(
     vt: &Vt,
+    base: usize,
     text: &'a str,
+    spans: &SpanMapper,
     styles: StyleChain<'a>,
     dir: Dir,
 ) -> ShapedText<'a> {
     let size = TextNode::size_in(styles);
     let mut ctx = ShapingContext {
         vt,
+        base,
+        spans,
         size,
         glyphs: vec![],
         used: vec![],
@@ -335,6 +365,7 @@ pub fn shape<'a>(
     track_and_space(&mut ctx);
 
     ShapedText {
+        base,
         text,
         dir,
         styles,
@@ -410,6 +441,7 @@ fn shape_segment<'a>(
         if info.glyph_id != 0 {
             // Add the glyph to the shaped output.
             // TODO: Don't ignore y_advance.
+            let (span, offset) = ctx.spans.span_at(ctx.base + cluster);
             ctx.glyphs.push(ShapedGlyph {
                 font: font.clone(),
                 glyph_id: info.glyph_id as u16,
@@ -419,6 +451,8 @@ fn shape_segment<'a>(
                 cluster: base + cluster,
                 safe_to_break: !info.unsafe_to_break(),
                 c: text[cluster..].chars().next().unwrap(),
+                span,
+                offset,
             });
         } else {
             // Determine the source text range for the tofu sequence.
@@ -478,15 +512,19 @@ fn shape_segment<'a>(
 fn shape_tofus(ctx: &mut ShapingContext, base: usize, text: &str, font: Font) {
     let x_advance = font.advance(0).unwrap_or_default();
     for (cluster, c) in text.char_indices() {
+        let cluster = base + cluster;
+        let (span, offset) = ctx.spans.span_at(ctx.base + cluster);
         ctx.glyphs.push(ShapedGlyph {
             font: font.clone(),
             glyph_id: 0,
             x_advance,
             x_offset: Em::zero(),
             y_offset: Em::zero(),
-            cluster: base + cluster,
+            cluster,
             safe_to_break: true,
             c,
+            span,
+            offset,
         });
     }
 }