SVGs with a loose "&" or HTML entities convert again

src/Filee.Engines/Vector/SvgRenderer.cs

@@ -56,10 +56,23 @@ public static Drawing Load(string path)
         svg.Settings.EnableJavaScript = false;
         svg.Settings.EnableExternalJavaScript = false;
         SKPicture? picture;
+        var uri = new Uri(Path.GetFullPath(path));
         try
         {
-            using var stream = new MemoryStream(bytes, writable: false);
-            picture = svg.Load(stream, null, new Uri(Path.GetFullPath(path)));
+            picture = Parse(svg, bytes, uri);
+        }
+        catch (Exception ex) when (ex is not OutOfMemoryException && SvgXmlRepair.FixAmpersands(bytes) is { } repaired)
+        {
+            // "A & B" or   in the text (common in generated SVGs): read it again with those escaped.
+            try
+            {
+                picture = Parse(svg, repaired, uri);
+            }
+            catch (Exception retry) when (retry is not OutOfMemoryException)
+            {
+                svg.Dispose();
+                throw new InvalidDataException("The SVG file could not be read: " + ex.Message, ex);
+            }
         }
         catch (Exception ex) when (ex is not OutOfMemoryException)
         {
@@ -75,6 +88,12 @@ public static Drawing Load(string path)
         return new Drawing(svg, picture, bounds);
     }
 
+    private static SKPicture? Parse(SKSvg svg, byte[] bytes, Uri baseUri)
+    {
+        using var stream = new MemoryStream(bytes, writable: false);
+        return svg.Load(stream, null, baseUri);
+    }
+
     /// <summary>The file's SVG text as bytes: SVGZ (or a gzipped .svg) is decompressed.</summary>
     public static byte[] ReadSvgBytes(string path)
     {

src/Filee.Engines/Vector/SvgXmlRepair.cs

@@ -0,0 +1,94 @@
+// Repairs the one mistake that makes many hand-written and AI-generated SVGs unreadable: an "&" that isn't an entity
+// ("A & B" in a <text>, a URL query in an href) and HTML entities XML doesn't know (&nbsp;, &copy;). Browsers forgive
+// some of it; the XML parser Svg.Skia uses stops at the first one (issue #120).
+
+using System.Net;
+using System.Text;
+using System.Text.RegularExpressions;
+
+namespace Filee.Engines.Vector;
+
+internal static partial class SvgXmlRepair
+{
+    /// <summary>The entities every XML document has; anything else must be declared in the DTD.</summary>
+    private static readonly HashSet<string> XmlEntities = ["amp", "lt", "gt", "quot", "apos"];
+
+    /// <summary>
+    /// Returns <paramref name="svg"/> with every "&amp;" that doesn't start a character reference or a known entity
+    /// escaped, and HTML named entities turned into character references. Entities declared in the document's DTD
+    /// (Illustrator's &amp;ns_svg;), the DOCTYPE, comments and CDATA sections are left alone. Null when nothing needed repair, or
+    /// when the bytes aren't UTF-8 (another declared encoding is read as it is).
+    /// </summary>
+    public static byte[]? FixAmpersands(byte[] svg)
+    {
+        if (svg.AsSpan().IndexOf((byte)'&') < 0)
+            return null;
+        string text;
+        try
+        {
+            text = new UTF8Encoding(false, throwOnInvalidBytes: true).GetString(svg);
+        }
+        catch (DecoderFallbackException)
+        {
+            return null;
+        }
+
+        var declared = DeclaredEntity().Matches(text).Select(m => m.Groups[1].Value).ToHashSet(StringComparer.Ordinal);
+        var result = new StringBuilder(text.Length + 64);
+        var changed = false;
+        var position = 0;
+        foreach (Match skip in Untouched().Matches(text))
+        {
+            changed |= Repair(text.AsSpan(position, skip.Index - position), declared, result);
+            result.Append(skip.Value);
+            position = skip.Index + skip.Length;
+        }
+        changed |= Repair(text.AsSpan(position), declared, result);
+        if (!changed)
+            return null;
+        return Encoding.UTF8.GetBytes(result.ToString()); // a byte order mark was decoded as U+FEFF and stays
+    }
+
+    /// <summary>Appends <paramref name="part"/> to <paramref name="result"/> repaired; true if anything changed.</summary>
+    private static bool Repair(ReadOnlySpan<char> part, HashSet<string> declared, StringBuilder result)
+    {
+        var changed = false;
+        var text = part.ToString();
+        var last = 0;
+        foreach (Match amp in Ampersand().Matches(text))
+        {
+            result.Append(text, last, amp.Index - last);
+            last = amp.Index + amp.Length;
+            var name = amp.Groups["name"];
+            if (amp.Groups["number"].Success || name.Success && (XmlEntities.Contains(name.Value) || declared.Contains(name.Value)))
+            {
+                result.Append(amp.Value);
+                continue;
+            }
+            changed = true;
+            if (name.Success && WebUtility.HtmlDecode(amp.Value) is { } decoded && decoded != amp.Value)
+            {
+                // An HTML entity: the same character as a reference XML reads.
+                foreach (var rune in decoded.EnumerateRunes())
+                    result.Append("&#").Append(rune.Value).Append(';');
+            }
+            else
+            {
+                // A lone "&", or an entity nobody defines: the "&" itself is text.
+                result.Append("&amp;").Append(amp.Value, 1, amp.Value.Length - 1);
+            }
+        }
+        result.Append(text, last, text.Length - last);
+        return changed;
+    }
+
+    // "&", then a character reference, a name with its ";", or neither (a lone "&").
+    [GeneratedRegex(@"&(?:(?<number>#[0-9]+;|#[xX][0-9A-Fa-f]+;)|(?<name>[A-Za-z_][A-Za-z0-9._-]*);)?")]
+    private static partial Regex Ampersand();
+
+    [GeneratedRegex(@"<!--.*?-->|<!\[CDATA\[.*?\]\]>|<!DOCTYPE[^\[>]*(?:\[.*?\])?\s*>", RegexOptions.Singleline)]
+    private static partial Regex Untouched();
+
+    [GeneratedRegex(@"<!ENTITY\s+([A-Za-z_][A-Za-z0-9._-]*)")]
+    private static partial Regex DeclaredEntity();
+}

tests/Filee.Engines.Tests/GraphicsTests.cs

@@ -321,6 +321,45 @@ public async Task Svg_to_png_is_rendered_large_enough_and_keeps_transparency()
             Assert.True(Pixel(flat, 3, 3).R > 245, "JPG gets the preset background");
     }
 
+    [Fact]
+    public async Task Svg_with_a_loose_ampersand_and_html_entities_still_converts()
+    {
+        // Issue #120: generated SVGs with "A & B" in a <text> stopped the XML parser ("parsing EntityName").
+        var dir = fx.NewFolder();
+        var svg = Write(dir, "gemini.svg", """
+            <svg xmlns="http://www.w3.org/2000/svg" width="200" height="100">
+              <rect width="200" height="100" fill="#ff0000"/>
+              <text x="10" y="50" font-size="20">Q&A &amp; R&nbsp;D &copy; 2026 &bogus;</text>
+            </svg>
+            """);
+
+        var job = await fx.ConvertAsync([svg], new Preset { TargetFormat = "png" });
+
+        AssertDone(job);
+        using var png = new MagickImage(job.Outputs.Single());
+        Assert.True(Pixel(png, 3, 3).R > 240, "the drawing was rendered");
+    }
+
+    [Theory]
+    [InlineData("<t>A & B</t>", "<t>A &amp; B</t>")]
+    [InlineData("<t>R&nbsp;D &copy;</t>", "<t>R&#160;D &#169;</t>")]
+    [InlineData("<t>&bogus; &copy</t>", "<t>&amp;bogus; &amp;copy</t>")]
+    [InlineData("<a href=\"x?a=1&b=2\"/>", "<a href=\"x?a=1&amp;b=2\"/>")]
+    public void Loose_ampersands_are_escaped_and_html_entities_become_references(string svg, string repaired) =>
+        Assert.Equal(repaired, Encoding.UTF8.GetString(SvgXmlRepair.FixAmpersands(Encoding.UTF8.GetBytes(svg))!));
+
+    [Theory]
+    [InlineData("<t>A &amp; B &lt; &#38; &#x26;</t>")]
+    [InlineData("<!DOCTYPE svg [<!ENTITY ns_svg \"http://www.w3.org/2000/svg\">]><svg xmlns=\"&ns_svg;\"/>")]
+    [InlineData("<t><!-- a & b --><![CDATA[x && y]]></t>")]
+    [InlineData("<t>no ampersand</t>")]
+    public void Valid_xml_is_left_as_it_is(string svg) =>
+        Assert.Null(SvgXmlRepair.FixAmpersands(Encoding.UTF8.GetBytes(svg)));
+
+    [Fact]
+    public void Text_in_another_encoding_is_not_touched() =>
+        Assert.Null(SvgXmlRepair.FixAmpersands(Encoding.Latin1.GetBytes("<t encoding=\"latin1\">café & co</t>")));
+
     [Fact]
     public async Task Svgz_round_trip()
     {