Support SVG sources (closes #68)

The fetcher accepts image/svg+xml; internal/magic detects SVG. orig of
an SVG at the orig size is the SVG unchanged, cached, with a
Content-Security-Policy allowing only its styles and data: images; at
another size, ErrSVGNotOriginalSize (400). Any other format renders it
at the least size covering the output, at most 8192 on a side, so it
stays sharp. Rendering refuses an SVG with a side over 8192 times the
other (ErrSVGTooThin, 415), and the orig size of one above 8192 on a
side (ErrSVGTooLargeForOrigSize, 400). The cache reads up to 1024 first
bytes to find a file's format. Tests pinning the SVG refusal now expect
SVG. Render time and memory are unbounded: P1 in TODO.md until the
owner decides on #68.

Model: opus-5-5

internal/magic/magic.go

@@ -20,6 +20,10 @@ var (
 // MinMagicBytes is the minimum number of bytes needed to detect format.
 const MinMagicBytes = 12
 
+// MaxMagicBytes is the most bytes DetectFormat looks at. An SVG's root
+// element must start within them, as libvips requires.
+const MaxMagicBytes = 1024
+
 // Magic byte signatures for supported formats.
 // These are effectively constants but Go doesn't support const slices.
 //
@@ -39,6 +43,12 @@ var (
 	}
 )
 
+// utf8BOM is the UTF-8 byte order mark, which a text file such as an SVG may
+// start with.
+//
+//nolint:gochecknoglobals // immutable lookup data
+var utf8BOM = []byte{0xEF, 0xBB, 0xBF}
+
 // WebP identifier appears at offset 8 after RIFF header.
 //
 //nolint:gochecknoglobals // immutable lookup data
@@ -53,7 +63,7 @@ var (
 )
 
 // DetectFormat detects the image format from magic bytes. Data in none of
-// the formats pixa reads, SVG included, is ErrUnknownFormat.
+// the formats pixa reads is ErrUnknownFormat.
 func DetectFormat(data []byte) (imagerequest.Format, error) {
 	if len(data) < MinMagicBytes {
 		return "", ErrNotEnoughData
@@ -96,9 +106,25 @@ func DetectFormat(data []byte) (imagerequest.Format, error) {
 		return imagerequest.FormatJXL, nil
 	}
 
+	if isSVG(data) {
+		return imagerequest.FormatSVG, nil
+	}
+
 	return "", ErrUnknownFormat
 }
 
+// isSVG reports whether data starts as an SVG does: with a "<", after an
+// optional UTF-8 byte order mark and white space, and with "<svg" within the
+// first MaxMagicBytes bytes. Before the root element an SVG may have an XML
+// declaration, comments and a DOCTYPE.
+func isSVG(data []byte) bool {
+	data = data[:min(len(data), MaxMagicBytes)]
+	text := bytes.TrimLeft(bytes.TrimPrefix(data, utf8BOM), " \t\r\n")
+
+	return bytes.HasPrefix(text, []byte("<")) &&
+		bytes.Contains(text, []byte("<svg"))
+}
+
 // ValidateMagicBytes detects the format of data from its magic bytes, and
 // returns it when it is the format the declared MIME type names. Data in
 // another format is ErrMagicByteMismatch.

internal/magic/magic_internal_test.go

@@ -47,9 +47,9 @@ func TestDetectFormat(t *testing.T) {
 		{name: "AVIF", data: avif, wantFormat: imagerequest.FormatAVIF},
 		{name: "AVIF sequence", data: avis, wantFormat: imagerequest.FormatAVIF},
 		{
-			name:    "SVG",
-			data:    []byte(`<svg xmlns="http://www.w3.org/2000/svg"></svg>`),
-			wantErr: ErrUnknownFormat,
+			name:       "SVG",
+			data:       []byte(`<svg xmlns="http://www.w3.org/2000/svg"></svg>`),
+			wantFormat: imagerequest.FormatSVG,
 		},
 		{
 			name:    "unknown format",

internal/magic/svg_internal_test.go

@@ -0,0 +1,85 @@
+package magic
+
+import (
+	"errors"
+	"strings"
+	"testing"
+
+	"sneak.berlin/go/pixa/internal/imagerequest"
+)
+
+// TestDetectFormat_SVG checks that DetectFormat names SVG text with what may
+// come before the root element, and does not name as SVG text that does not
+// start with "<" or has no "<svg" within its first MaxMagicBytes bytes.
+func TestDetectFormat_SVG(t *testing.T) {
+	t.Parallel()
+
+	const root = `<svg xmlns="http://www.w3.org/2000/svg" width="4" height="4"/>`
+
+	tests := []struct {
+		name    string
+		data    string
+		wantErr error
+	}{
+		{name: "root element only", data: root},
+		{name: "XML declaration", data: `<?xml version="1.0"?>` + "\n" + root},
+		{name: "byte order mark and white space", data: "\xEF\xBB\xBF \n\t" + root},
+		{name: "comment", data: "<!-- a drawing -->" + root},
+		{
+			name: "DOCTYPE",
+			data: `<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" ` +
+				`"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">` + root,
+		},
+		{
+			name:    "root element after the first bytes",
+			data:    "<!--" + strings.Repeat(" ", MaxMagicBytes) + "-->" + root,
+			wantErr: ErrUnknownFormat,
+		},
+		{name: "text naming <svg", data: "draw it with " + root, wantErr: ErrUnknownFormat},
+		{name: "HTML", data: "<!DOCTYPE html><html></html>", wantErr: ErrUnknownFormat},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Parallel()
+
+			got, err := DetectFormat([]byte(tt.data))
+			t.Logf("DetectFormat() = %q, %v", got, err)
+
+			if tt.wantErr != nil {
+				if !errors.Is(err, tt.wantErr) {
+					t.Errorf("DetectFormat() error = %v, want %v", err, tt.wantErr)
+				}
+
+				return
+			}
+
+			if err != nil || got != imagerequest.FormatSVG {
+				t.Errorf("DetectFormat() = %q, %v; want %q", got, err,
+					imagerequest.FormatSVG)
+			}
+		})
+	}
+}
+
+// TestValidateMagicBytes_SVG checks that SVG data is SVG for the
+// Content-Type image/svg+xml, with a charset too, and a mismatch for another.
+func TestValidateMagicBytes_SVG(t *testing.T) {
+	t.Parallel()
+
+	data := []byte(`<svg xmlns="http://www.w3.org/2000/svg" width="4" height="4"/>`)
+
+	for _, declared := range []string{"image/svg+xml", "image/svg+xml; charset=utf-8"} {
+		got, err := ValidateMagicBytes(data, declared)
+		if err != nil || got != imagerequest.FormatSVG {
+			t.Errorf("ValidateMagicBytes(%q) = %q, %v; want %q", declared, got,
+				err, imagerequest.FormatSVG)
+		}
+	}
+
+	_, err := ValidateMagicBytes(data, "image/png")
+	if !errors.Is(err, ErrMagicByteMismatch) {
+		t.Errorf("ValidateMagicBytes(image/png) error = %v, want %v", err,
+			ErrMagicByteMismatch)
+	}
+}