Add internal URL type (#5074)

4 files changed, 94 insertions, 46 deletions

diff --git a/crates/typst-ide/src/jump.rs b/crates/typst-ide/src/jump.rs
index d7f2e1d2..b798defa 100644
--- a/crates/typst-ide/src/jump.rs
+++ b/crates/typst-ide/src/jump.rs
@@ -1,8 +1,7 @@
 use std::num::NonZeroUsize;
 
-use ecow::EcoString;
 use typst::layout::{Frame, FrameItem, Point, Position, Size};
-use typst::model::{Destination, Document};
+use typst::model::{Destination, Document, Url};
 use typst::syntax::{FileId, LinkedNode, Side, Source, Span, SyntaxKind};
 use typst::visualize::Geometry;
 use typst::World;
@@ -13,7 +12,7 @@ pub enum Jump {
     /// Jump to a position in a source file.
     Source(FileId, usize),
     /// Jump to an external URL.
-    Url(EcoString),
+    Url(Url),
     /// Jump to a point on a page.
     Position(Position),
 }
diff --git a/crates/typst/src/eval/markup.rs b/crates/typst/src/eval/markup.rs
index a735bb8e..42fede1c 100644
--- a/crates/typst/src/eval/markup.rs
+++ b/crates/typst/src/eval/markup.rs
@@ -1,4 +1,4 @@
-use crate::diag::{warning, SourceResult};
+use crate::diag::{warning, At, SourceResult};
 use crate::eval::{Eval, Vm};
 use crate::foundations::{
     Content, Label, NativeElement, Repr, Smart, Unlabellable, Value,
@@ -6,7 +6,7 @@ use crate::foundations::{
 use crate::math::EquationElem;
 use crate::model::{
     EmphElem, EnumItem, HeadingElem, LinkElem, ListItem, ParbreakElem, RefElem,
-    StrongElem, Supplement, TermItem,
+    StrongElem, Supplement, TermItem, Url,
 };
 use crate::symbols::Symbol;
 use crate::syntax::ast::{self, AstNode};
@@ -195,7 +195,8 @@ impl Eval for ast::Link<'_> {
     type Output = Content;
 
     fn eval(self, _: &mut Vm) -> SourceResult<Self::Output> {
-        Ok(LinkElem::from_url(self.get().clone()).pack())
+        let url = Url::new(self.get().clone()).at(self.span())?;
+        Ok(LinkElem::from_url(url).pack())
     }
 }
 
diff --git a/crates/typst/src/model/bibliography.rs b/crates/typst/src/model/bibliography.rs
index 11454804..d6b81a20 100644
--- a/crates/typst/src/model/bibliography.rs
+++ b/crates/typst/src/model/bibliography.rs
@@ -34,6 +34,7 @@ use crate::layout::{
 };
 use crate::model::{
     CitationForm, CiteGroup, Destination, FootnoteElem, HeadingElem, LinkElem, ParElem,
+    Url,
 };
 
 use crate::syntax::{Span, Spanned};
@@ -741,8 +742,8 @@ impl<'a> Generator<'a> {
 
     /// Displays hayagriva's output as content for the citations and references.
     fn display(&mut self, rendered: &hayagriva::Rendered) -> StrResult<Works> {
-        let citations = self.display_citations(rendered);
-        let references = self.display_references(rendered);
+        let citations = self.display_citations(rendered)?;
+        let references = self.display_references(rendered)?;
         let hanging_indent =
             rendered.bibliography.as_ref().is_some_and(|b| b.hanging_indent);
         Ok(Works { citations, references, hanging_indent })
@@ -752,7 +753,7 @@ impl<'a> Generator<'a> {
     fn display_citations(
         &mut self,
         rendered: &hayagriva::Rendered,
-    ) -> HashMap<Location, SourceResult<Content>> {
+    ) -> StrResult<HashMap<Location, SourceResult<Content>>> {
         // Determine for each citation key where in the bibliography it is,
         // so that we can link there.
         let mut links = HashMap::new();
@@ -779,7 +780,7 @@ impl<'a> Generator<'a> {
                 Content::empty()
             } else {
                 let mut content =
-                    renderer.display_elem_children(&citation.citation, &mut None);
+                    renderer.display_elem_children(&citation.citation, &mut None)?;
 
                 if info.footnote {
                     content = FootnoteElem::with_content(content).pack();
@@ -791,15 +792,16 @@ impl<'a> Generator<'a> {
             output.insert(info.location, Ok(content));
         }
 
-        output
+        Ok(output)
     }
 
     /// Display the bibliography references.
+    #[allow(clippy::type_complexity)]
     fn display_references(
         &self,
         rendered: &hayagriva::Rendered,
-    ) -> Option<Vec<(Option<Content>, Content)>> {
-        let rendered = rendered.bibliography.as_ref()?;
+    ) -> StrResult<Option<Vec<(Option<Content>, Content)>>> {
+        let Some(rendered) = &rendered.bibliography else { return Ok(None) };
 
         // Determine for each citation key where it first occurred, so that we
         // can link there.
@@ -829,18 +831,22 @@ impl<'a> Generator<'a> {
             let backlink = location.variant(k + 1);
 
             // Render the first field.
-            let mut prefix = item.first_field.as_ref().map(|elem| {
-                let mut content = renderer.display_elem_child(elem, &mut None);
-                if let Some(location) = first_occurrences.get(item.key.as_str()) {
-                    let dest = Destination::Location(*location);
-                    content = content.linked(dest);
-                }
-                content
-            });
+            let mut prefix = item
+                .first_field
+                .as_ref()
+                .map(|elem| {
+                    let mut content = renderer.display_elem_child(elem, &mut None)?;
+                    if let Some(location) = first_occurrences.get(item.key.as_str()) {
+                        let dest = Destination::Location(*location);
+                        content = content.linked(dest);
+                    }
+                    StrResult::Ok(content)
+                })
+                .transpose()?;
 
             // Render the main reference content.
             let mut reference =
-                renderer.display_elem_children(&item.content, &mut prefix);
+                renderer.display_elem_children(&item.content, &mut prefix)?;
 
             // Attach a backlink to either the prefix or the reference so that
             // we can link to the bibliography entry.
@@ -849,7 +855,7 @@ impl<'a> Generator<'a> {
             output.push((prefix, reference));
         }
 
-        Some(output)
+        Ok(Some(output))
     }
 }
 
@@ -874,10 +880,14 @@ impl ElemRenderer<'_> {
         &self,
         elems: &hayagriva::ElemChildren,
         prefix: &mut Option<Content>,
-    ) -> Content {
-        Content::sequence(
-            elems.0.iter().map(|elem| self.display_elem_child(elem, prefix)),
-        )
+    ) -> StrResult<Content> {
+        Ok(Content::sequence(
+            elems
+                .0
+                .iter()
+                .map(|elem| self.display_elem_child(elem, prefix))
+                .collect::<StrResult<Vec<_>>>()?,
+        ))
     }
 
     /// Display a rendered hayagriva element.
@@ -885,16 +895,16 @@ impl ElemRenderer<'_> {
         &self,
         elem: &hayagriva::ElemChild,
         prefix: &mut Option<Content>,
-    ) -> Content {
-        match elem {
+    ) -> StrResult<Content> {
+        Ok(match elem {
             hayagriva::ElemChild::Text(formatted) => self.display_formatted(formatted),
-            hayagriva::ElemChild::Elem(elem) => self.display_elem(elem, prefix),
+            hayagriva::ElemChild::Elem(elem) => self.display_elem(elem, prefix)?,
             hayagriva::ElemChild::Markup(markup) => self.display_math(markup),
-            hayagriva::ElemChild::Link { text, url } => self.display_link(text, url),
+            hayagriva::ElemChild::Link { text, url } => self.display_link(text, url)?,
             hayagriva::ElemChild::Transparent { cite_idx, format } => {
                 self.display_transparent(*cite_idx, format)
             }
-        }
+        })
     }
 
     /// Display a block-level element.
@@ -902,7 +912,7 @@ impl ElemRenderer<'_> {
         &self,
         elem: &hayagriva::Elem,
         prefix: &mut Option<Content>,
-    ) -> Content {
+    ) -> StrResult<Content> {
         use citationberg::Display;
 
         let block_level = matches!(elem.display, Some(Display::Block | Display::Indent));
@@ -911,7 +921,7 @@ impl ElemRenderer<'_> {
         let mut content = self.display_elem_children(
             &elem.children,
             if block_level { &mut suf_prefix } else { prefix },
-        );
+        )?;
 
         if let Some(prefix) = suf_prefix {
             const COLUMN_GUTTER: Em = Em::new(0.65);
@@ -941,7 +951,7 @@ impl ElemRenderer<'_> {
             }
             Some(Display::LeftMargin) => {
                 *prefix.get_or_insert_with(Default::default) += content;
-                return Content::empty();
+                return Ok(Content::empty());
             }
             _ => {}
         }
@@ -953,7 +963,7 @@ impl ElemRenderer<'_> {
             }
         }
 
-        content
+        Ok(content)
     }
 
     /// Display math.
@@ -964,11 +974,11 @@ impl ElemRenderer<'_> {
     }
 
     /// Display a link.
-    fn display_link(&self, text: &hayagriva::Formatted, url: &str) -> Content {
-        let dest = Destination::Url(url.into());
-        LinkElem::new(dest.into(), self.display_formatted(text))
+    fn display_link(&self, text: &hayagriva::Formatted, url: &str) -> StrResult<Content> {
+        let dest = Destination::Url(Url::new(url)?);
+        Ok(LinkElem::new(dest.into(), self.display_formatted(text))
             .pack()
-            .spanned(self.span)
+            .spanned(self.span))
     }
 
     /// Display transparent pass-through content.
diff --git a/crates/typst/src/model/link.rs b/crates/typst/src/model/link.rs
index 107b0d9a..b583a6fd 100644
--- a/crates/typst/src/model/link.rs
+++ b/crates/typst/src/model/link.rs
@@ -1,7 +1,9 @@
+use std::ops::Deref;
+
 use ecow::{eco_format, EcoString};
 use smallvec::SmallVec;
 
-use crate::diag::{At, SourceResult};
+use crate::diag::{bail, At, SourceResult, StrResult};
 use crate::engine::Engine;
 use crate::foundations::{
     cast, elem, Content, Label, Packed, Repr, Show, Smart, StyleChain,
@@ -89,7 +91,7 @@ pub struct LinkElem {
 
 impl LinkElem {
     /// Create a link element from a URL with its bare text.
-    pub fn from_url(url: EcoString) -> Self {
+    pub fn from_url(url: Url) -> Self {
         let body = body_from_url(&url);
         Self::new(LinkTarget::Dest(Destination::Url(url)), body)
     }
@@ -112,13 +114,13 @@ impl Show for Packed<LinkElem> {
     }
 }
 
-fn body_from_url(url: &EcoString) -> Content {
+fn body_from_url(url: &Url) -> Content {
     let mut text = url.as_str();
     for prefix in ["mailto:", "tel:"] {
         text = text.trim_start_matches(prefix);
     }
     let shorter = text.len() < url.len();
-    TextElem::packed(if shorter { text.into() } else { url.clone() })
+    TextElem::packed(if shorter { text.into() } else { (**url).clone() })
 }
 
 /// A target where a link can go.
@@ -148,13 +150,15 @@ impl From<Destination> for LinkTarget {
 #[derive(Debug, Clone, Eq, PartialEq, Hash)]
 pub enum Destination {
     /// A link to a URL.
-    Url(EcoString),
+    Url(Url),
     /// A link to a point on a page.
     Position(Position),
     /// An unresolved link to a location in the document.
     Location(Location),
 }
 
+impl Destination {}
+
 impl Repr for Destination {
     fn repr(&self) -> EcoString {
         eco_format!("{self:?}")
@@ -168,7 +172,41 @@ cast! {
         Self::Position(v) => v.into_value(),
         Self::Location(v) => v.into_value(),
     },
-    v: EcoString => Self::Url(v),
+    v: Url => Self::Url(v),
     v: Position => Self::Position(v),
     v: Location => Self::Location(v),
 }
+
+/// A uniform resource locator with a maximum length.
+#[derive(Debug, Clone, Eq, PartialEq, Ord, PartialOrd, Hash)]
+pub struct Url(EcoString);
+
+impl Url {
+    /// Create an URL from a string, checking the maximum length.
+    pub fn new(url: impl Into<EcoString>) -> StrResult<Self> {
+        let url = url.into();
+        if url.len() > 8000 {
+            bail!("URL is too long")
+        }
+        Ok(Self(url))
+    }
+
+    /// Extract the underlying [`EcoString`].
+    pub fn into_inner(self) -> EcoString {
+        self.0
+    }
+}
+
+impl Deref for Url {
+    type Target = EcoString;
+
+    fn deref(&self) -> &Self::Target {
+        &self.0
+    }
+}
+
+cast! {
+    Url,
+    self => self.0.into_value(),
+    v: EcoString => Self::new(v)?,
+}