Support smartquotes in HTML export (#6710)

Co-authored-by: Malo <57839069+MDLC01@users.noreply.github.com>

crates/typst-html/src/convert.rs

@@ -5,20 +5,24 @@ use typst_library::foundations::{Content, StyleChain, Target, TargetElem};
 use typst_library::introspection::{SplitLocator, TagElem};
 use typst_library::layout::{Abs, Axes, Region, Size};
 use typst_library::routines::Pair;
-use typst_library::text::{LinebreakElem, SmartQuoteElem, SpaceElem, TextElem};
+use typst_library::text::{
+    LinebreakElem, SmartQuoteElem, SmartQuoter, SmartQuotes, SpaceElem, TextElem,
+    is_default_ignorable,
+};
 
-use crate::fragment::html_fragment;
+use crate::fragment::{html_block_fragment, html_inline_fragment};
 use crate::{FrameElem, HtmlElem, HtmlElement, HtmlFrame, HtmlNode, tag};
 
 /// Converts realized content into HTML nodes.
 pub fn convert_to_nodes<'a>(
     engine: &mut Engine,
     locator: &mut SplitLocator,
+    quoter: &mut SmartQuoter,
     children: impl IntoIterator<Item = Pair<'a>>,
 ) -> SourceResult<EcoVec<HtmlNode>> {
     let mut output = EcoVec::new();
     for (child, styles) in children {
-        handle(engine, child, locator, styles, &mut output)?;
+        handle(engine, child, locator, styles, quoter, &mut output)?;
     }
     Ok(output)
 }
@@ -29,6 +33,7 @@ fn handle(
     child: &Content,
     locator: &mut SplitLocator,
     styles: StyleChain,
+    quoter: &mut SmartQuoter,
     output: &mut EcoVec<HtmlNode>,
 ) -> SourceResult<()> {
     if let Some(elem) = child.to_packed::<TagElem>() {
@@ -36,7 +41,22 @@ fn handle(
     } else if let Some(elem) = child.to_packed::<HtmlElem>() {
         let mut children = EcoVec::new();
         if let Some(body) = elem.body.get_ref(styles) {
-            children = html_fragment(engine, body, locator.next(&elem.span()), styles)?;
+            if tag::is_block_by_default(elem.tag) {
+                children = html_block_fragment(
+                    engine,
+                    body,
+                    locator.next(&elem.span()),
+                    styles,
+                )?;
+
+                // Block-level elements reset the smart quoting state. This part
+                // is unfortunately untested as it's currently not possible to
+                // create inline-level content next to block-level content
+                // without a paragraph automatically appearing.
+                *quoter = SmartQuoter::new();
+            } else {
+                children = html_inline_fragment(engine, body, locator, quoter, styles)?;
+            }
         }
         let element = HtmlElement {
             tag: elem.tag,
@@ -57,10 +77,20 @@ fn handle(
     } else if let Some(elem) = child.to_packed::<LinebreakElem>() {
         output.push(HtmlElement::new(tag::br).spanned(elem.span()).into());
     } else if let Some(elem) = child.to_packed::<SmartQuoteElem>() {
-        output.push(HtmlNode::text(
+        let double = elem.double.get(styles);
-            if elem.double.get(styles) { '"' } else { '\'' },
+        if elem.enabled.get(styles) {
-            child.span(),
+            let before = last_char(output);
-        ));
+            let quotes = SmartQuotes::get(
+                elem.quotes.get_ref(styles),
+                styles.get(TextElem::lang),
+                styles.get(TextElem::region),
+                elem.alternative.get(styles),
+            );
+            let quote = quoter.quote(before, &quotes, double);
+            output.push(HtmlNode::text(quote, child.span()));
+        } else {
+            output.push(HtmlNode::text(if double { '"' } else { '\'' }, child.span()));
+        }
     } else if let Some(elem) = child.to_packed::<FrameElem>() {
         let locator = locator.next(&elem.span());
         let style = TargetElem::target.set(Target::Paged).wrap();
@@ -82,6 +112,20 @@ fn handle(
     Ok(())
 }
 
+/// Returns the last non-default ignorable character from the passed nodes.
+fn last_char(nodes: &[HtmlNode]) -> Option<char> {
+    for node in nodes.iter().rev() {
+        if let Some(c) = match node {
+            HtmlNode::Text(s, _) => s.chars().rev().find(|&c| !is_default_ignorable(c)),
+            HtmlNode::Element(e) => last_char(&e.children),
+            _ => None,
+        } {
+            return Some(c);
+        }
+    }
+    None
+}
+
 /// Checks whether the given element is an inline-level HTML element.
 pub fn is_inline(elem: &Content) -> bool {
     elem.to_packed::<HtmlElem>()

crates/typst-html/src/document.rs

@@ -13,6 +13,7 @@ use typst_library::introspection::{
 use typst_library::layout::{Point, Position, Transform};
 use typst_library::model::DocumentInfo;
 use typst_library::routines::{Arenas, RealizationKind, Routines};
+use typst_library::text::SmartQuoter;
 use typst_syntax::Span;
 use typst_utils::NonZeroExt;
 
@@ -85,6 +86,7 @@ fn html_document_impl(
     let output = crate::convert::convert_to_nodes(
         &mut engine,
         &mut locator,
+        &mut SmartQuoter::new(),
         children.iter().copied(),
     )?;
 

crates/typst-html/src/fragment.rs

@@ -3,22 +3,24 @@ use ecow::EcoVec;
 use typst_library::diag::{At, SourceResult};
 use typst_library::engine::{Engine, Route, Sink, Traced};
 use typst_library::foundations::{Content, StyleChain};
-use typst_library::introspection::{Introspector, Locator, LocatorLink};
+use typst_library::introspection::{Introspector, Locator, LocatorLink, SplitLocator};
 
 use typst_library::World;
-use typst_library::routines::{Arenas, FragmentKind, RealizationKind, Routines};
+use typst_library::routines::{Arenas, FragmentKind, Pair, RealizationKind, Routines};
+use typst_library::text::SmartQuoter;
 
 use crate::HtmlNode;
 
-/// Produce HTML nodes from content.
+/// Produces HTML nodes from content contained in an HTML element that is
-#[typst_macros::time(name = "html fragment")]
+/// block-level by default.
-pub fn html_fragment(
+#[typst_macros::time(name = "html block fragment")]
+pub fn html_block_fragment(
     engine: &mut Engine,
     content: &Content,
     locator: Locator,
     styles: StyleChain,
 ) -> SourceResult<EcoVec<HtmlNode>> {
-    html_fragment_impl(
+    html_block_fragment_impl(
         engine.routines,
         engine.world,
         engine.introspector,
@@ -34,7 +36,7 @@ pub fn html_fragment(
 /// The cached, internal implementation of [`html_fragment`].
 #[comemo::memoize]
 #[allow(clippy::too_many_arguments)]
-fn html_fragment_impl(
+fn html_block_fragment_impl(
     routines: &Routines,
     world: Tracked<dyn World + '_>,
     introspector: Tracked<Introspector>,
@@ -59,19 +61,65 @@ fn html_fragment_impl(
     engine.route.check_html_depth().at(content.span())?;
 
     let arenas = Arenas::default();
-    let children = (engine.routines.realize)(
+    let children = realize_fragment(&mut engine, &mut locator, &arenas, content, styles)?;
-        // No need to know about the `FragmentKind` because we handle both
+    crate::convert::convert_to_nodes(
-        // uniformly.
+        &mut engine,
+        &mut locator,
+        &mut SmartQuoter::new(),
+        children.iter().copied(),
+    )
+}
+
+/// Produces HTML nodes from content contained in an HTML element that is
+/// inline-level by default.
+///
+/// The difference to block-level content is that inline-level content has
+/// shared smartquoting state with surrounding inline-level content. This
+/// requires mutable state, which is at odds with memoization. However, the
+/// caching granularity would be unnecessarily high anyway if every single
+/// fragment was cached, so this works out pretty well together.
+#[typst_macros::time(name = "html inline fragment")]
+pub fn html_inline_fragment(
+    engine: &mut Engine,
+    content: &Content,
+    locator: &mut SplitLocator,
+    quoter: &mut SmartQuoter,
+    styles: StyleChain,
+) -> SourceResult<EcoVec<HtmlNode>> {
+    engine.route.increase();
+    engine.route.check_html_depth().at(content.span())?;
+
+    let arenas = Arenas::default();
+    let children = realize_fragment(engine, locator, &arenas, content, styles)?;
+    let result = crate::convert::convert_to_nodes(
+        engine,
+        locator,
+        quoter,
+        children.iter().copied(),
+    );
+
+    engine.route.decrease();
+    result
+}
+
+/// Realizes the body of an HTML fragment.
+fn realize_fragment<'a>(
+    engine: &mut Engine,
+    locator: &mut SplitLocator,
+    arenas: &'a Arenas,
+    content: &'a Content,
+    styles: StyleChain<'a>,
+) -> SourceResult<Vec<Pair<'a>>> {
+    (engine.routines.realize)(
         RealizationKind::HtmlFragment {
+            // We ignore the `FragmentKind` because we handle both uniformly.
             kind: &mut FragmentKind::Block,
             is_inline: crate::convert::is_inline,
         },
-        &mut engine,
+        engine,
-        &mut locator,
+        locator,
-        &arenas,
+        arenas,
         content,
         styles,
-    )?;
+    )
-
-    crate::convert::convert_to_nodes(&mut engine, &mut locator, children.iter().copied())
 }

tests/ref/html/par-semantic-html.html

@@ -6,8 +6,8 @@
   </head>
   <body>
     <h2>Heading is no paragraph</h2>
-    <p>I'm a paragraph.</p>
+    <p>I’m a paragraph.</p>
-    <div>I'm not.</div>
+    <div>I’m not.</div>
     <div>
       <p>We are two.</p>
       <p>So we are paragraphs.</p>

tests/ref/html/quote-nesting-html.html

@@ -5,6 +5,6 @@
     <meta name="viewport" content="width=device-width, initial-scale=1">
   </head>
   <body>
-    <p>When you said that “he surely meant that ‘she intended to say “I'm sorry”’”, I was quite confused.</p>
+    <p>When you said that “he surely meant that ‘she intended to say “I’m sorry”’”, I was quite confused.</p>
   </body>
 </html>

tests/ref/html/smartquote-inline-block.html

@@ -0,0 +1,13 @@
+<!DOCTYPE html>
+<html>
+  <head>
+    <meta charset="utf-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1">
+  </head>
+  <body>
+    <p>Applies across <span>“inline-level</span> elements”.</p>
+    <p>Does not apply across</p>
+    <div>“block-level</div>
+    <p>elements“.</p>
+  </body>
+</html>

tests/ref/html/smartquote-nesting-twice.html

@@ -0,0 +1,11 @@
+<!DOCTYPE html>
+<html>
+  <head>
+    <meta charset="utf-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1">
+  </head>
+  <body>
+    <p>When you said <em>that “he</em> surely meant that ‘she intended to say “I’m sorry”’”, I was quite confused.</p>
+    <p>‘<span style="display: inline-block">box</span>’</p>
+  </body>
+</html>

tests/ref/html/smartquotes-html.html

@@ -0,0 +1,11 @@
+<!DOCTYPE html>
+<html>
+  <head>
+    <meta charset="utf-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1">
+  </head>
+  <body>
+    <p>When you said that “he surely meant that ‘she intended to say “I’m sorry”’”, I was quite confused.</p>
+    <p>‘<span style="display: inline-block;">box</span>’</p>
+  </body>
+</html>

tests/suite/text/smartquote.typ

@@ -113,6 +113,16 @@ Some people's thought on this would be #[#set smartquote(enabled: false); "stran
 "'test' statement" \
 "statement 'test'"
 
+--- smartquote-nesting-twice render html ---
+When you said _that "he_ surely meant that 'she intended to say "I'm sorry"'", I was quite confused.
+
+'#box[box]'
+
+--- smartquote-inline-block html ---
+Applies across #html.span["inline-level] elements".
+
+Does not apply across #html.div["block-level] elements".
+
 --- smartquote-with-embedding-chars ---
 #set text(lang: "fr")
 "#"\u{202A}"bonjour#"\u{202C}"" \