XML reading

src/diag.rs

@@ -3,6 +3,7 @@
 use std::fmt::{self, Display, Formatter};
 use std::io;
 use std::path::{Path, PathBuf};
+use std::str::Utf8Error;
 use std::string::FromUtf8Error;
 
 use comemo::Tracked;
@@ -191,6 +192,12 @@ impl Display for FileError {
     }
 }
 
+impl From<Utf8Error> for FileError {
+    fn from(_: Utf8Error) -> Self {
+        Self::InvalidUtf8
+    }
+}
+
 impl From<FromUtf8Error> for FileError {
     fn from(_: FromUtf8Error) -> Self {
         Self::InvalidUtf8
@@ -202,3 +209,33 @@ impl From<FileError> for String {
         error.to_string()
     }
 }
+
+/// Format a user-facing error message for an XML-like file format.
+pub fn format_xml_like_error(format: &str, error: roxmltree::Error) -> String {
+    match error {
+        roxmltree::Error::UnexpectedCloseTag { expected, actual, pos } => {
+            format!(
+                "failed to parse {format}: found closing tag '{actual}' \
+                 instead of '{expected}' in line {}",
+                pos.row
+            )
+        }
+        roxmltree::Error::UnknownEntityReference(entity, pos) => {
+            format!(
+                "failed to parse {format}: unknown entity '{entity}' in line {}",
+                pos.row
+            )
+        }
+        roxmltree::Error::DuplicatedAttribute(attr, pos) => {
+            format!(
+                "failed to parse {format}: duplicate attribute '{attr}' in line {}",
+                pos.row
+            )
+        }
+        roxmltree::Error::NoRootNode => {
+            format!("failed to parse {format}: missing root node")
+        }
+        roxmltree::Error::SizeLimit => "file is too large".into(),
+        _ => format!("failed to parse {format}"),
+    }
+}

src/image.rs

@@ -2,7 +2,7 @@
 
 use std::io;
 
-use crate::diag::StrResult;
+use crate::diag::{format_xml_like_error, StrResult};
 use crate::util::Buffer;
 
 /// A raster or vector image.
@@ -161,31 +161,6 @@ fn format_usvg_error(error: usvg::Error) -> String {
         usvg::Error::InvalidSize => {
             "failed to parse svg: width, height, or viewbox is invalid".into()
         }
-        usvg::Error::ParsingFailed(error) => match error {
-            roxmltree::Error::UnexpectedCloseTag { expected, actual, pos } => {
-                format!(
-                    "failed to parse svg: found closing tag '{actual}' \
-                     instead of '{expected}' in line {}",
-                    pos.row
-                )
-            }
-            roxmltree::Error::UnknownEntityReference(entity, pos) => {
-                format!(
-                    "failed to parse svg: unknown entity '{entity}' in line {}",
-                    pos.row
-                )
-            }
-            roxmltree::Error::DuplicatedAttribute(attr, pos) => {
-                format!(
-                    "failed to parse svg: duplicate attribute '{attr}' in line {}",
-                    pos.row
-                )
-            }
-            roxmltree::Error::NoRootNode => {
-                "failed to parse svg: missing root node".into()
-            }
-            roxmltree::Error::SizeLimit => "file is too large".into(),
-            _ => "failed to parse svg".into(),
-        },
+        usvg::Error::ParsingFailed(error) => format_xml_like_error("svg", error),
     }
 }

src/library/mod.rs

@@ -98,6 +98,7 @@ pub fn new() -> Scope {
     std.def_fn("lorem", utility::lorem);
     std.def_fn("csv", utility::csv);
     std.def_fn("json", utility::json);
+    std.def_fn("xml", utility::xml);
 
     // Predefined colors.
     std.define("black", Color::BLACK);

src/library/prelude.rs

@@ -9,10 +9,12 @@ pub use std::sync::Arc;
 pub use comemo::Tracked;
 pub use typst_macros::node;
 
-pub use crate::diag::{with_alternative, At, SourceError, SourceResult, StrResult};
+pub use crate::diag::{
+    with_alternative, At, FileError, FileResult, SourceError, SourceResult, StrResult,
+};
 pub use crate::eval::{
     Arg, Args, Array, Cast, Dict, Dynamic, Func, Node, RawAlign, RawLength, RawStroke,
-    Scope, Smart, Value, Vm,
+    Scope, Smart, Str, Value, Vm,
 };
 pub use crate::frame::*;
 pub use crate::geom::*;

src/library/utility/data.rs

@@ -1,5 +1,6 @@
 use std::fmt::Write;
 
+use crate::diag::format_xml_like_error;
 use crate::library::prelude::*;
 
 /// Read structured data from a CSV file.
@@ -84,3 +85,47 @@ fn format_json_error(error: serde_json::Error) -> String {
         error.line()
     )
 }
+
+/// Read structured data from an XML file.
+pub fn xml(vm: &mut Vm, args: &mut Args) -> SourceResult<Value> {
+    let Spanned { v: path, span } =
+        args.expect::<Spanned<EcoString>>("path to xml file")?;
+
+    let path = vm.locate(&path).at(span)?;
+    let data = vm.world.file(&path).at(span)?;
+    let text = std::str::from_utf8(&data).map_err(FileError::from).at(span)?;
+
+    let document = roxmltree::Document::parse(text).map_err(format_xml_error).at(span)?;
+
+    Ok(convert_xml(document.root()))
+}
+
+/// Convert an XML node to a Typst value.
+fn convert_xml(node: roxmltree::Node) -> Value {
+    if node.is_text() {
+        return Value::Str(node.text().unwrap_or_default().into());
+    }
+
+    let children: Array = node.children().map(convert_xml).collect();
+    if node.is_root() {
+        return Value::Array(children);
+    }
+
+    let tag: Str = node.tag_name().name().into();
+    let attrs: Dict = node
+        .attributes()
+        .iter()
+        .map(|attr| (attr.name().into(), attr.value().into()))
+        .collect();
+
+    Value::Dict(dict! {
+        "tag" => tag,
+        "attrs" => attrs,
+        "children" => children,
+    })
+}
+
+/// Format the user-facing XML error message.
+fn format_xml_error(error: roxmltree::Error) -> String {
+    format_xml_like_error("xml file", error)
+}

tests/res/bad.xml

@@ -0,0 +1,3 @@
+<data>
+  <hello name="hi">1
+</data>

tests/res/data.xml

@@ -0,0 +1,7 @@
+<data>
+  <hello name="hi">1</hello>
+  <data>
+    <hello>World</hello>
+    <hello>World</hello>
+  </data>
+</data>

tests/typ/utility/data.typ

@@ -27,3 +27,32 @@
 ---
 // Error: 7-22 failed to parse json file: syntax error in line 3
 #json("/res/bad.json")
+
+---
+// Test reading XML data.
+#let data = xml("/res/data.xml")
+#test(data, ((
+  tag: "data",
+  attrs: (:),
+  children: (
+    "\n  ",
+    (tag: "hello", attrs: (name: "hi"), children: ("1",)),
+    "\n  ",
+    (
+      tag: "data",
+      attrs: (:),
+      children: (
+        "\n    ",
+        (tag: "hello", attrs: (:), children: ("World",)),
+        "\n    ",
+        (tag: "hello", attrs: (:), children: ("World",)),
+        "\n  ",
+      ),
+    ),
+    "\n",
+  ),
+),))
+
+---
+// Error: 6-20 failed to parse xml file: found closing tag 'data' instead of 'hello' in line 3
+#xml("/res/bad.xml")