More robust SVG auto-detection (#5878)

Cargo.lock

@@ -2966,6 +2966,7 @@ dependencies = [
  "kamadak-exif",
  "kurbo",
  "lipsum",
+ "memchr",
  "palette",
  "phf",
  "png",

Cargo.toml

@@ -73,6 +73,7 @@ kamadak-exif = "0.6"
 kurbo = "0.11"
 libfuzzer-sys = "0.4"
 lipsum = "0.9"
+memchr = "2"
 miniz_oxide = "0.8"
 native-tls = "0.2"
 notify = "8"

crates/typst-library/Cargo.toml

@@ -38,6 +38,7 @@ indexmap = { workspace = true }
 kamadak-exif = { workspace = true }
 kurbo = { workspace = true }
 lipsum = { workspace = true }
+memchr = { workspace = true }
 palette = { workspace = true }
 phf = { workspace = true }
 png = { workspace = true }

crates/typst-library/src/visualize/image/mod.rs

@@ -398,8 +398,7 @@ impl ImageFormat {
             return Some(Self::Raster(RasterFormat::Exchange(format)));
         }
 
-        // SVG or compressed SVG.
-        if data.starts_with(b"<svg") || data.starts_with(&[0x1f, 0x8b]) {
+        if is_svg(data) {
             return Some(Self::Vector(VectorFormat::Svg));
         }
 
@@ -407,6 +406,21 @@ impl ImageFormat {
     }
 }
 
+/// Checks whether the data looks like an SVG or a compressed SVG.
+fn is_svg(data: &[u8]) -> bool {
+    // Check for the gzip magic bytes. This check is perhaps a bit too
+    // permissive as other formats than SVGZ could use gzip.
+    if data.starts_with(&[0x1f, 0x8b]) {
+        return true;
+    }
+
+    // If the first 2048 bytes contain the SVG namespace declaration, we assume
+    // that it's an SVG. Note that, if the SVG does not contain a namespace
+    // declaration, usvg will reject it.
+    let head = &data[..data.len().min(2048)];
+    memchr::memmem::find(head, b"http://www.w3.org/2000/svg").is_some()
+}
+
 /// A vector graphics format.
 #[derive(Debug, Copy, Clone, Eq, PartialEq, Hash, Cast)]
 pub enum VectorFormat {

docs/changelog/0.13.0.md

@@ -99,8 +99,7 @@ description: Changes slated to appear in Typst 0.13.0
 - Fixed interaction of clipping and outset on [`box`] and [`block`]
 - Fixed panic with [`path`] of infinite length
 - Fixed non-solid (e.g. tiling) text fills in clipped blocks
-- Auto-detection of image formats from a raw buffer now has basic support for
-  SVGs
+- Auto-detection of image formats from a raw buffer now has support for SVGs
 
 ## Scripting
 - Functions that accept [file paths]($syntax/#paths) now also accept raw

tests/suite/visualize/image.typ

@@ -65,6 +65,17 @@ A #box(image("/assets/images/tiger.jpg", height: 1cm, width: 80%)) B
   caption: [Bilingual text]
 )
 
+--- image-svg-auto-detection ---
+#image(bytes(
+  ```
+  <?xml version="1.0" encoding="utf-8"?>
+  <!-- An SVG -->
+  <svg width="200" height="150" xmlns="http://www.w3.org/2000/svg">
+    <rect fill="red" stroke="black" x="25" y="25" width="150" height="100"/>
+  </svg>
+  ```.text
+))
+
 --- image-pixmap-rgb8 ---
 #image(
   bytes((
@@ -152,8 +163,8 @@ A #box(image("/assets/images/tiger.jpg", height: 1cm, width: 80%)) B
 #image("path/does/not/exist")
 
 --- image-bad-format ---
-// Error: 2-22 unknown image format
-#image("./image.typ")
+// Error: 2-37 unknown image format
+#image("/assets/plugins/hello.wasm")
 
 --- image-bad-svg ---
 // Error: 2-33 failed to parse SVG (found closing tag 'g' instead of 'style' in line 4)