test: add tests for language tag attributes

Cargo.lock

@@ -3326,7 +3326,6 @@ dependencies = [
  "clap",
  "comemo 0.4.0",
  "ecow",
- "krilla",
  "oxipng",
  "parking_lot",
  "rayon",

crates/typst-pdf/src/convert.rs

@@ -9,7 +9,8 @@ use krilla::geom::PathBuilder;
 use krilla::page::{PageLabel, PageSettings};
 use krilla::pdf::PdfError;
 use krilla::surface::Surface;
-use krilla::tagging::{TagId, TagTree};
+use krilla::tagging::TagId;
+use krilla::tagging::fmt::Output;
 use krilla::{Document, SerializeSettings};
 use krilla_svg::render_svg_glyph;
 use typst_library::diag::{SourceDiagnostic, SourceResult, bail, error};
@@ -19,7 +20,7 @@ use typst_library::layout::{
     Abs, Frame, FrameItem, GroupItem, PagedDocument, Size, Transform,
 };
 use typst_library::model::HeadingElem;
-use typst_library::text::Font;
+use typst_library::text::{Font, Lang};
 use typst_library::visualize::{Geometry, Paint};
 use typst_syntax::Span;
 
@@ -55,10 +56,18 @@ pub fn convert(
 pub fn tag_tree(
     typst_document: &PagedDocument,
     options: &PdfOptions,
-) -> SourceResult<TagTree> {
+) -> SourceResult<String> {
     let (mut document, mut gc) = setup(typst_document, options);
     convert_pages(&mut gc, &mut document)?;
-    Ok(gc.tags.build_tree())
+    let mut tree = if let Some(lang) = gc.tags.doc_lang
+        && lang != Lang::ENGLISH
+    {
+        format!("lang: \"{}\"\n---\n", lang.as_str())
+    } else {
+        String::new()
+    };
+    gc.tags.build_tree().output(&mut tree).unwrap();
+    Ok(tree)
 }
 
 fn setup<'a>(

crates/typst-pdf/src/lib.rs

@@ -18,7 +18,6 @@ pub use self::metadata::{Timestamp, Timezone};
 use std::fmt::{self, Debug, Formatter};
 
 use ecow::eco_format;
-use krilla::tagging::TagTree;
 use serde::{Deserialize, Serialize};
 use typst_library::diag::{SourceResult, StrResult, bail};
 use typst_library::foundations::Smart;
@@ -34,7 +33,10 @@ pub fn pdf(document: &PagedDocument, options: &PdfOptions) -> SourceResult<Vec<u
 
 /// Generate the document tag tree and display it in a human readable form.
 #[doc(hidden)]
-pub fn pdf_tags(document: &PagedDocument, options: &PdfOptions) -> SourceResult<TagTree> {
+pub fn pdf_tags(
+    document: &PagedDocument,
+    options: &PdfOptions,
+) -> SourceResult<String> {
     convert::tag_tree(document, options)
 }
 

tests/Cargo.toml

@@ -25,7 +25,6 @@ default = [
     "typst-render",
     "typst-svg",
     "typst-svg",
-    "krilla",
 ]
 
 [dependencies]
@@ -40,7 +39,6 @@ typst-library = { workspace = true, optional = true }
 typst-pdf = { workspace = true, optional = true }
 typst-render = { workspace = true, optional = true }
 typst-svg = { workspace = true, optional = true }
-krilla = { workspace = true, optional = true }
 clap = { workspace = true }
 comemo = { workspace = true }
 ecow = { workspace = true }

tests/ref/pdftags/lang-tags-pars-basic.yml

@@ -0,0 +1,13 @@
+lang: "uk"
+---
+- Tag: P
+  /K:
+    - Content: page=0 mcid=0
+- Tag: P
+  /Lang: "sr"
+  /K:
+    - Content: page=0 mcid=1
+- Tag: P
+  /Lang: "be"
+  /K:
+    - Content: page=0 mcid=2

tests/ref/pdftags/lang-tags-propagation.yml

@@ -0,0 +1,46 @@
+lang: "nl"
+---
+- Tag: P
+  /K:
+    - Content: page=0 mcid=0
+- Tag: L
+  /Lang: "de"
+  /Numbering: Circle
+  /K:
+    - Tag: LI
+      /K:
+        - Tag: Lbl
+          /K:
+            - Content: page=0 mcid=1
+        - Tag: LBody
+          /K:
+            - Tag: P
+              /K:
+                - Content: page=0 mcid=2
+    - Tag: L
+      /Numbering: Circle
+      /K:
+        - Tag: LI
+          /K:
+            - Tag: Lbl
+              /K:
+                - Content: page=0 mcid=3
+            - Tag: LBody
+              /K:
+                - Content: page=0 mcid=4
+        - Tag: LI
+          /K:
+            - Tag: Lbl
+              /K:
+                - Content: page=0 mcid=5
+            - Tag: LBody
+              /K:
+                - Content: page=0 mcid=6
+    - Tag: LI
+      /K:
+        - Tag: Lbl
+          /K:
+            - Content: page=0 mcid=7
+        - Tag: LBody
+          /K:
+            - Content: page=0 mcid=8

tests/ref/pdftags/text-lang.yml

@@ -0,0 +1,8 @@
+lang: "zh"
+---
+- Tag: H1
+  /T: "目录"
+  /K:
+    - Content: page=0 mcid=0
+    - Content: page=0 mcid=1
+- Tag: TOC

tests/src/run.rs

@@ -3,8 +3,6 @@ use std::ops::Range;
 use std::path::PathBuf;
 
 use ecow::eco_vec;
-use krilla::tagging::TagTree;
-use krilla::tagging::fmt::Output;
 use tiny_skia as sk;
 use typst::diag::{SourceDiagnostic, SourceResult, Warned};
 use typst::layout::{Abs, Frame, FrameItem, PagedDocument, Transform};
@@ -76,7 +74,7 @@ impl<'a> Runner<'a> {
             self.run_test::<HtmlDocument>();
         }
         if pdftags {
-            self.run_test::<TagTree>();
+            self.run_test::<Pdftags>();
         }
 
         self.handle_not_emitted();
@@ -518,7 +516,9 @@ impl OutputType for HtmlDocument {
     }
 }
 
-impl OutputType for TagTree {
+struct Pdftags(String);
+
+impl OutputType for Pdftags {
     type Live = String;
 
     fn live_path(name: &str) -> PathBuf {
@@ -537,16 +537,16 @@ impl OutputType for TagTree {
         };
         let mut options = PdfOptions::default();
         options.standards = PdfStandards::new(&[PdfStandard::Ua_1]).unwrap();
-        let output = typst_pdf::pdf_tags(&doc, &options);
+        let output = typst_pdf::pdf_tags(&doc, &options).map(Pdftags);
         Warned { warnings, output }
     }
 
     fn is_skippable(&self) -> Result<bool, ()> {
-        Ok(self.children.is_empty())
+        Ok(self.0.is_empty())
     }
 
     fn make_live(&self) -> SourceResult<Self::Live> {
-        Ok(self.display().to_string())
+        Ok(self.0.clone())
     }
 
     fn save_live(&self, name: &str, live: &Self::Live) {

tests/suite/pdftags/lang.typ

@@ -0,0 +1,19 @@
+--- lang-tags-pars-basic pdftags ---
+#set text(lang: "uk")
+Par 1.
+
+#set text(lang: "sr")
+Par 2.
+
+#set text(lang: "be")
+Par 3.
+
+--- lang-tags-propagation pdftags ---
+#set text(lang: "nl")
+A paragraph.
+
+// language attributes are propagated to the parent (L) tag
+- #text(lang: "de", "a")
+  - #text(lang: "de", "b")
+  - #text(lang: "de", "c")
+- #text(lang: "de", "d")

tests/suite/text/lang.typ

@@ -1,6 +1,6 @@
 // Test setting the document language.
 
---- text-lang ---
+--- text-lang render pdftags ---
 // without any region
 #set text(font: "Noto Serif CJK TC", lang: "zh")
 #outline()