WIP [no ci] tag table headers

crates/typst-library/src/model/outline.rs

@@ -273,6 +273,7 @@ impl Show for Packed<OutlineElem> {
         let depth = self.depth(styles).unwrap_or(NonZeroUsize::MAX);
 
         // Build the outline entries.
+        let mut entries = vec![];
         for elem in elems {
             let Some(outlinable) = elem.with::<dyn Outlinable>() else {
                 bail!(span, "cannot outline {}", elem.func().name());
@@ -281,10 +282,13 @@ impl Show for Packed<OutlineElem> {
             let level = outlinable.level();
             if outlinable.outlined() && level <= depth {
                 let entry = OutlineEntry::new(level, elem);
-                seq.push(entry.pack().spanned(span));
+                entries.push(entry.pack().spanned(span));
             }
         }
 
+        // Wrap the entries into a marker for pdf tagging.
+        seq.push(OutlineBody::new(Content::sequence(entries)).pack());
+
         Ok(Content::sequence(seq))
     }
 }
@@ -307,6 +311,19 @@ impl LocalName for Packed<OutlineElem> {
     const KEY: &'static str = "outline";
 }
 
+/// Only used to mark
+#[elem(Locatable, Show)]
+pub struct OutlineBody {
+    #[required]
+    body: Content,
+}
+
+impl Show for Packed<OutlineBody> {
+    fn show(&self, _: &mut Engine, _: StyleChain) -> SourceResult<Content> {
+        Ok(self.body.clone())
+    }
+}
+
 /// Defines how an outline is indented.
 #[derive(Debug, Clone, PartialEq, Hash)]
 pub enum OutlineIndent {

crates/typst-library/src/model/table.rs

@@ -2,6 +2,7 @@ use std::num::{NonZeroU32, NonZeroUsize};
 use std::sync::Arc;
 
 use ecow::EcoString;
+use typst_macros::Cast;
 use typst_utils::NonZeroExt;
 
 use crate::diag::{bail, HintedStrResult, HintedString, SourceResult};
@@ -361,15 +362,40 @@ fn show_cellgrid_html(grid: CellGrid, styles: StyleChain) -> Content {
 
 impl Show for Packed<TableElem> {
     fn show(&self, engine: &mut Engine, styles: StyleChain) -> SourceResult<Content> {
-        Ok(if TargetElem::target_in(styles).is_html() {
+        let elem = if TargetElem::target_in(styles).is_html() {
             // TODO: This is a hack, it is not clear whether the locator is actually used by HTML.
             // How can we find out whether locator is actually used?
             let locator = Locator::root();
             show_cellgrid_html(table_to_cellgrid(self, engine, locator, styles)?, styles)
         } else {
-            BlockElem::multi_layouter(self.clone(), engine.routines.layout_table).pack()
+            let children = self
+                .children
+                .iter()
+                .map(|c| match c.clone() {
+                    TableChild::Header(header) => {
+                        let mut header = header.unpack();
+                        header.children = header
+                            .children
+                            .into_iter()
+                            .map(|item| match item {
+                                TableItem::Cell(cell) => {
+                                    let mut cell = cell.unpack();
+                                    cell.push_header_scope(Smart::Custom(TableHeaderScope::Column));
+                                    TableItem::Cell(Packed::new(cell))
                                 }
-        .spanned(self.span()))
+                                item => item,
+                            })
+                            .collect();
+                        TableChild::Header(Packed::new(header))
+                    }
+                    child => child,
+                })
+                .collect();
+            let mut table = self.clone().unpack();
+            table.children = children;
+            BlockElem::multi_layouter(Packed::new(table), engine.routines.layout_table).pack()
+        };
+        Ok(elem.spanned(self.span()))
     }
 }
 
@@ -810,6 +836,9 @@ pub struct TableCell {
     #[fold]
     pub stroke: Sides<Option<Option<Arc<Stroke>>>>,
 
+    // TODO: feature gate
+    pub header_scope: Smart<TableHeaderScope>,
+
     /// Whether rows spanned by this cell can be placed in different pages.
     /// When equal to `{auto}`, a cell spanning only fixed-size rows is
     /// unbreakable, while a cell spanning at least one `{auto}`-sized row is
@@ -847,3 +876,10 @@ impl From<Content> for TableCell {
         value.unpack::<Self>().unwrap_or_else(Self::new)
     }
 }
+
+#[derive(Debug, Copy, Clone, Eq, PartialEq, Hash, Cast)]
+pub enum TableHeaderScope {
+    Both,
+    Column,
+    Row,
+}

crates/typst-pdf/src/convert.rs

@@ -171,14 +171,14 @@ impl State {
 /// Context needed for converting a single frame.
 pub(crate) struct FrameContext {
     states: Vec<State>,
-    pub(crate) link_annotations: HashMap<tags::LinkId, LinkAnnotation>,
+    link_annotations: Vec<LinkAnnotation>,
 }
 
 impl FrameContext {
     pub(crate) fn new(size: Size) -> Self {
         Self {
             states: vec![State::new(size)],
-            link_annotations: HashMap::new(),
+            link_annotations: Vec::new(),
         }
     }
 
@@ -197,6 +197,20 @@ impl FrameContext {
     pub(crate) fn state_mut(&mut self) -> &mut State {
         self.states.last_mut().unwrap()
     }
+
+    pub(crate) fn get_link_annotation(
+        &mut self,
+        link_id: tags::LinkId,
+    ) -> Option<&mut LinkAnnotation> {
+        self.link_annotations
+            .iter_mut()
+            .rev()
+            .find(|annot| annot.id == link_id)
+    }
+
+    pub(crate) fn push_link_annotation(&mut self, annotation: LinkAnnotation) {
+        self.link_annotations.push(annotation);
+    }
 }
 
 /// Globally needed context for converting a typst document.

crates/typst-pdf/src/link.rs

@@ -1,18 +1,18 @@
-use std::collections::hash_map::Entry;
-
 use ecow::EcoString;
 use krilla::action::{Action, LinkAction};
 use krilla::annotation::Target;
+use krilla::configure::Validator;
 use krilla::destination::XyzDestination;
 use krilla::geom as kg;
 use typst_library::layout::{Abs, Point, Position, Size};
 use typst_library::model::Destination;
 
 use crate::convert::{FrameContext, GlobalContext};
-use crate::tags::{Placeholder, StackEntryKind, TagNode};
+use crate::tags::{self, Placeholder, StackEntryKind, TagNode};
 use crate::util::{AbsExt, PointExt};
 
 pub(crate) struct LinkAnnotation {
+    pub(crate) id: tags::LinkId,
     pub(crate) placeholder: Placeholder,
     pub(crate) alt: Option<String>,
     pub(crate) rect: kg::Rect,
@@ -50,7 +50,7 @@ pub(crate) fn handle_link(
     };
 
     let entry = gc.tags.stack.last_mut().expect("a link parent");
-    let StackEntryKind::Link(link_id, link) = &entry.kind else {
+    let StackEntryKind::Link(link_id, ref link) = entry.kind else {
         unreachable!("expected a link parent")
     };
     let alt = link.alt.as_ref().map(EcoString::to_string);
@@ -58,18 +58,21 @@ pub(crate) fn handle_link(
     let rect = to_rect(fc, size);
     let quadpoints = quadpoints(rect);
 
-    match fc.link_annotations.entry(*link_id) {
-        Entry::Occupied(occupied) => {
-            // Update the bounding box and add the quadpoints of an existing link annotation.
-            let annotation = occupied.into_mut();
+    // Unfortunately quadpoints still aren't well supported by most PDF readers,
+    // even by acrobat. Which is understandable since they were only introduced
+    // in PDF 1.6 (2005) /s
+    let should_use_quadpoints = gc.options.standards.config.validator() == Validator::UA1;
+    match fc.get_link_annotation(link_id) {
+        Some(annotation) if should_use_quadpoints => {
+            // Update the bounding box and add the quadpoints to an existing link annotation.
             annotation.rect = bounding_rect(annotation.rect, rect);
             annotation.quad_points.extend_from_slice(&quadpoints);
         }
-        Entry::Vacant(vacant) => {
+        _ => {
             let placeholder = gc.tags.reserve_placeholder();
             gc.tags.push(TagNode::Placeholder(placeholder));
-
-            vacant.insert(LinkAnnotation {
+            fc.push_link_annotation(LinkAnnotation {
+                id: link_id,
                 placeholder,
                 rect,
                 quad_points: quadpoints.to_vec(),

crates/typst-pdf/src/tags.rs

@@ -1,19 +1,18 @@
 use std::cell::OnceCell;
-use std::collections::HashMap;
 
 use ecow::EcoString;
 use krilla::page::Page;
 use krilla::surface::Surface;
 use krilla::tagging::{
     ArtifactType, ContentTag, Identifier, Node, SpanTag, TableCellSpan, TableDataCell,
-    TableHeaderCell, TableHeaderScope, Tag, TagBuilder, TagGroup, TagKind, TagTree,
+    TableHeaderCell, Tag, TagBuilder, TagGroup, TagKind, TagTree,
 };
-use typst_library::foundations::{Content, LinkMarker, Packed, StyleChain};
+use typst_library::foundations::{Content, LinkMarker, Packed, StyleChain, Smart};
 use typst_library::introspection::Location;
 use typst_library::layout::RepeatElem;
 use typst_library::model::{
-    Destination, FigureCaption, FigureElem, HeadingElem, Outlinable, OutlineElem,
-    OutlineEntry, TableCell, TableElem,
+    Destination, FigureCaption, FigureElem, HeadingElem, Outlinable, OutlineBody,
+    OutlineEntry, TableCell, TableElem, TableHeaderScope,
 };
 use typst_library::pdf::{ArtifactElem, ArtifactKind, PdfTagElem, PdfTagKind};
 use typst_library::visualize::ImageElem;
@@ -41,11 +40,11 @@ pub(crate) struct StackEntry {
 
 pub(crate) enum StackEntryKind {
     Standard(Tag),
-    Link(LinkId, Packed<LinkMarker>),
     Outline(OutlineCtx),
     OutlineEntry(Packed<OutlineEntry>),
     Table(TableCtx),
     TableCell(Packed<TableCell>),
+    Link(LinkId, Packed<LinkMarker>),
 }
 
 impl StackEntryKind {
@@ -142,10 +141,14 @@ impl TableCtx {
 
         // TODO: possibly set internal field on TableCell when resolving
         // the cell grid.
-        let is_header = false;
+        let header_scope = cell.header_scope(StyleChain::default());
         let span = TableCellSpan { rows: rowspan as i32, cols: colspan as i32 };
-        let tag = if is_header {
-            let scope = TableHeaderScope::Column; // TODO
+        let tag = if let Smart::Custom(scope) = header_scope {
+            let scope = match scope {
+                TableHeaderScope::Both => krilla::tagging::TableHeaderScope::Both,
+                TableHeaderScope::Column => krilla::tagging::TableHeaderScope::Column,
+                TableHeaderScope::Row => krilla::tagging::TableHeaderScope::Row,
+            };
             TagKind::TH(TableHeaderCell::new(scope).with_span(span))
         } else {
             TagKind::TD(TableDataCell::new().with_span(span))
@@ -168,14 +171,37 @@ impl TableCtx {
     fn build_table(self, mut nodes: Vec<TagNode>) -> Vec<TagNode> {
         // Table layouting ensures that there are no overlapping cells, and that
         // any gaps left by the user are filled with empty cells.
+        let mut row_chunk = Vec::new();
+        let mut is_header_chunk = false;
         for row in self.rows.into_iter() {
-            let mut row_nodes = Vec::new();
-            for (_, tag, nodes) in row.into_iter().flatten() {
-                row_nodes.push(TagNode::Group(tag, nodes));
+            let mut is_header_row = true;
+            let row_nodes = row
+                .into_iter()
+                .flatten()
+                .map(|(cell, tag, nodes)| {
+                    is_header_row &= cell.header_scope(StyleChain::default()).is_custom();
+                    TagNode::Group(tag, nodes)
+                })
+                .collect();
+
+            let row = TagNode::Group(TagKind::TR.into(), row_nodes);
+            if row_chunk.is_empty() {
+                is_header_chunk = is_header_row;
+                row_chunk.push(row);
+            } else if is_header_chunk == is_header_row {
+                row_chunk.push(row);
+            } else {
+                let tag = if is_header_chunk { TagKind::THead } else { TagKind::TBody };
+                nodes.push(TagNode::Group(tag.into(), std::mem::take(&mut row_chunk)));
+
+                is_header_chunk = is_header_row;
+                row_chunk.push(row);
+            }
         }
 
-            // TODO: generate `THead`, `TBody`, and `TFoot`
-            nodes.push(TagNode::Group(TagKind::TR.into(), row_nodes));
+        if !row_chunk.is_empty() {
+            let tag = if is_header_chunk { TagKind::THead } else { TagKind::TBody };
+            nodes.push(TagNode::Group(tag.into(), row_chunk));
         }
 
         nodes
@@ -335,10 +361,11 @@ fn start_content<'a, 'b>(
 pub(crate) fn add_annotations(
     gc: &mut GlobalContext,
     page: &mut Page,
-    annotations: HashMap<LinkId, LinkAnnotation>,
+    annotations: Vec<LinkAnnotation>,
 ) {
-    for annotation in annotations.into_values() {
-        let LinkAnnotation { placeholder, alt, rect, quad_points, target } = annotation;
+    for annotation in annotations.into_iter() {
+        let LinkAnnotation { id: _, placeholder, alt, rect, quad_points, target } =
+            annotation;
         let annot = krilla::annotation::Annotation::new_link(
             krilla::annotation::LinkAnnotation::new(rect, Some(quad_points), target),
             alt,
@@ -383,7 +410,7 @@ pub(crate) fn handle_start(gc: &mut GlobalContext, elem: &Content) {
             // TODO: when targeting PDF 2.0 headings `> 6` are supported
             _ => TagKind::H6(Some(name)).into(),
         }
-    } else if let Some(_) = elem.to_packed::<OutlineElem>() {
+    } else if let Some(_) = elem.to_packed::<OutlineBody>() {
         push_stack(gc, loc, StackEntryKind::Outline(OutlineCtx::new()));
         return;
     } else if let Some(entry) = elem.to_packed::<OutlineEntry>() {
@@ -409,16 +436,16 @@ pub(crate) fn handle_start(gc: &mut GlobalContext, elem: &Content) {
         }
     } else if let Some(_) = elem.to_packed::<FigureCaption>() {
         TagKind::Caption.into()
-    } else if let Some(link) = elem.to_packed::<LinkMarker>() {
-        let link_id = gc.tags.next_link_id();
-        push_stack(gc, loc, StackEntryKind::Link(link_id, link.clone()));
-        return;
     } else if let Some(table) = elem.to_packed::<TableElem>() {
         push_stack(gc, loc, StackEntryKind::Table(TableCtx::new(table.clone())));
         return;
     } else if let Some(cell) = elem.to_packed::<TableCell>() {
         push_stack(gc, loc, StackEntryKind::TableCell(cell.clone()));
         return;
+    } else if let Some(link) = elem.to_packed::<LinkMarker>() {
+        let link_id = gc.tags.next_link_id();
+        push_stack(gc, loc, StackEntryKind::Link(link_id, link.clone()));
+        return;
     } else {
         return;
     };
@@ -448,16 +475,6 @@ pub(crate) fn handle_end(gc: &mut GlobalContext, loc: Location) {
 
     let node = match entry.kind {
         StackEntryKind::Standard(tag) => TagNode::Group(tag, entry.nodes),
-        StackEntryKind::Link(_, link) => {
-            let alt = link.alt.as_ref().map(EcoString::to_string);
-            let tag = TagKind::Link.with_alt_text(alt);
-            let mut node = TagNode::Group(tag, entry.nodes);
-            // Wrap link in reference tag, if it's not a url.
-            if let Destination::Position(_) | Destination::Location(_) = link.dest {
-                node = TagNode::Group(TagKind::Reference.into(), vec![node]);
-            }
-            node
-        }
         StackEntryKind::Outline(ctx) => {
             let nodes = ctx.build_outline(entry.nodes);
             TagNode::Group(TagKind::TOC.into(), nodes)
@@ -487,6 +504,16 @@ pub(crate) fn handle_end(gc: &mut GlobalContext, loc: Location) {
 
             return;
         }
+        StackEntryKind::Link(_, link) => {
+            let alt = link.alt.as_ref().map(EcoString::to_string);
+            let tag = TagKind::Link.with_alt_text(alt);
+            let mut node = TagNode::Group(tag, entry.nodes);
+            // Wrap link in reference tag, if it's not a url.
+            if let Destination::Position(_) | Destination::Location(_) = link.dest {
+                node = TagNode::Group(TagKind::Reference.into(), vec![node]);
+            }
+            node
+        }
     };
 
     gc.tags.push(node);