Skip to content

nullean/mermaider

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

26 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Mermaider

Mermaider

Render Mermaid diagrams to SVG in pure .NET.
No browser. No DOM. No JavaScript runtime. AOT-ready.

NuGet CI


Why Mermaider?

Most .NET packages for Mermaid fall into one of two camps: DSL-only libraries that help you build Mermaid markup but can't render it, or browser wrappers that shell out to Chrome, Puppeteer, or a Node.js process to produce SVGs. Both have trade-offs—the first gives you a string you still can't display, the second drags in a JavaScript runtime with all its latency, memory overhead, and deployment complexity.

Mermaider is neither. It is a complete parser, lightweight layout engine and renderer implemented entirely in .NET. Hand it a Mermaid string, get an SVG back. No interop, no child processes, no headless browsers.

Pure .NET parsing and rendering

Mermaider parses Mermaid's text DSL and renders SVG output using only managed .NET code. There is no dependency on JavaScript, Chromium, or any external process. This means deterministic output, no cold-start penalty, and trivial deployment—just a NuGet reference.

Built-in layout engine

Graph-based diagrams (flowchart, state, class, ER) need a layout algorithm to position nodes and route edges. Other diagram types (pie, quadrant, timeline, gitgraph, radar, treemap, venn, mindmap) use purpose-built layout arithmetic directly in their renderers. Rather than depending on an external engine, Mermaider ships its own lightweight Sugiyama layout engine with zero dependencies.

During development, Microsoft MSAGL (Automatic Graph Layout) was evaluated as the layout backend. MSAGL is a capable research-grade library, but it carries baggage from a different era of .NET: high allocations (~554 KB for a 6-node flowchart), WPF-era BinaryFormatter usage, and trim/AOT warnings that make it unsuitable for modern deployment targets.

The built-in engine is purpose-built for the small-to-medium directed graphs Mermaid produces:

Phase MSAGL Built-in Sugiyama Improvement
Layout only 247 µs / 558 KB 3.4 µs / 16 KB 73× faster, 35× less memory
End-to-end render 351 µs / 586 KB 24 µs / 46 KB 15× faster, 13× less memory

If you still want MSAGL for its higher-fidelity edge routing on complex graphs, install the optional Mermaider.Layout.Msagl package (see below).

Native AOT

Every public API is compatible with .NET Native AOT. The CI pipeline publishes and invokes a native binary on Linux, macOS, and Windows to prove it. No reflection, no runtime code generation, no surprises.

Quick Start

dotnet add package Mermaider
using Mermaider;

var svg = MermaidRenderer.RenderSvg("""
    graph TD
      A[Start] --> B{Decision}
      B -->|Yes| C[OK]
      B -->|No| D[End]
    """);

Supported Diagrams

Flowchart

MermaidRenderer.RenderSvg("""
    graph TD
      A[Start] --> B{Decision}
      B -->|Yes| C[OK]
      B -->|No| D[Cancel]
      C --> E[End]
      D --> E
    """);

Flowchart

Sequence

MermaidRenderer.RenderSvg("""
    sequenceDiagram
      participant A as Alice
      participant B as Bob
      A->>B: Hello Bob!
      B-->>A: Hi Alice!
      A->>B: How are you?
      B-->>A: Great, thanks!
    """);

Sequence diagram

State

MermaidRenderer.RenderSvg("""
    stateDiagram-v2
      [*] --> Idle
      Idle --> Processing : submit
      Processing --> Success : ok
      Processing --> Failed : error
      Success --> [*]
      Failed --> Idle : retry
    """);

State diagram

Class

MermaidRenderer.RenderSvg("""
    classDiagram
      class Animal {
        <<abstract>>
        +String name
        +eat() void
      }
      class Dog { +bark() void }
      class Cat { +purr() void }
      Animal <|-- Dog
      Animal <|-- Cat
    """);

Class diagram

ER (Entity-Relationship)

MermaidRenderer.RenderSvg("""
    erDiagram
      CUSTOMER ||--o{ ORDER : places
      ORDER ||--|{ LINE_ITEM : contains
      CUSTOMER {
        string name PK
        string email UK
      }
      ORDER {
        int id PK
        date created
      }
    """);

ER diagram

Pie Chart

MermaidRenderer.RenderSvg("""
    pie
    title Pet Adoption
    "Dogs" : 386
    "Cats" : 85
    "Rats" : 15
    """);

Pie chart

Quadrant Chart

MermaidRenderer.RenderSvg("""
    quadrantChart
    title Priority Matrix
    x-axis Low Effort --> High Effort
    y-axis Low Impact --> High Impact
    quadrant-1 Do First
    quadrant-2 Schedule
    quadrant-3 Delegate
    quadrant-4 Eliminate
    Feature A: [0.8, 0.9]
    Feature B: [0.2, 0.3]
    Feature C: [0.6, 0.4]
    """);

Quadrant chart

Timeline

MermaidRenderer.RenderSvg("""
    timeline
    title History of Social Media
    section Early Days
    2002 : LinkedIn
    2004 : Facebook : Google
    section Modern Era
    2010 : Instagram
    2019 : TikTok
    """);

Timeline diagram

GitGraph

MermaidRenderer.RenderSvg("""
    gitGraph
    commit id: "init"
    commit id: "feat-1"
    branch develop
    checkout develop
    commit id: "dev-1"
    commit id: "dev-2" tag: "v0.1"
    checkout main
    merge develop id: "merge-1"
    commit id: "release" type: HIGHLIGHT tag: "v1.0"
    """);

GitGraph

Radar Chart

MermaidRenderer.RenderSvg("""
    radar-beta
    title Skills Assessment
    axis Design, Frontend, Backend, DevOps, Testing
    curve c1["Team A"]{4, 3, 5, 2, 4}
    curve c2["Team B"]{3, 5, 2, 4, 3}
    max 5
    graticule polygon
    """);

Radar chart

Treemap

MermaidRenderer.RenderSvg("""
    treemap-beta
    "Engineering": 50
    "Marketing": 25
    "Sales": 15
    "Support": 10
    """);

Treemap

Venn Diagram

MermaidRenderer.RenderSvg("""
    venn-beta
    set A["Frontend"]
    set B["Backend"]
    set C["DevOps"]
    union A, B["Full Stack"]
    union B, C["SRE"]
    """);

Venn diagram

Mindmap

MermaidRenderer.RenderSvg("""
    mindmap
      ((Project))
        (Planning)
          Requirements
          Timeline
        [Development]
          Frontend
          Backend
        {{Testing}}
          Unit Tests
          Integration
    """);

Mindmap

Theming

Every diagram derives its palette from just two colors—background and foreground—using color-mix() CSS functions embedded in the SVG. Override individual roles for richer themes:

var svg = MermaidRenderer.RenderSvg(input, new RenderOptions
{
    Bg = "#1E1E2E",
    Fg = "#CDD6F4",
    Accent = "#CBA6F7",    // arrow heads, highlights
    Muted  = "#6C7086",    // secondary text, labels
});

Because the SVG uses CSS custom properties, themes switch live without re-rendering—just update the --bg / --fg properties on the root <svg> element.

Strict Mode

When you embed user-authored Mermaid in a product, you typically want uniform styling controlled by your design system—not arbitrary colors injected via classDef or style directives.

Strict mode:

  • Rejects classDef and style directives at parse time (throws MermaidParseException)
  • Enforces a pre-approved class allowlist with theme-aware colors
  • Generates @media (prefers-color-scheme: dark) CSS for automatic light/dark switching
  • Auto-derives dark mode colors by inverting HSL lightness (or use explicit overrides)
var svg = MermaidRenderer.RenderSvg(input, new RenderOptions
{
    Strict = new StrictModeOptions
    {
        AllowedClasses =
        [
            new DiagramClass
            {
                Name = "ok",
                Fill = "#D4EDDA", Stroke = "#28A745", Color = "#155724",
            },
            new DiagramClass
            {
                Name = "warn",
                Fill = "#FFF3CD", Stroke = "#FFC107", Color = "#856404",
            },
            new DiagramClass { Name = "custom-highlight" },
        ],
        RejectUnknownClasses = true,
        Sanitize = SvgSanitizeMode.Strip,
    }
});

Nodes reference classes via Mermaid's ::: shorthand or class directive:

graph TD
  A[Healthy]:::ok --> B[Warning]:::warn --> C[Custom]:::custom-highlight

SVG Sanitization

A standalone, general-purpose SVG sanitizer is included—useful beyond Mermaid for any untrusted SVG content.

It enforces element and attribute allowlists, and always blocks the main XSS vectors regardless of the allowlist: <script>, <foreignObject>, on* event handlers, href/xlink:href with javascript: URIs.

var result = SvgSanitizer.Sanitize(untrustedSvg);

if (result.HasViolations)
    Console.WriteLine($"Stripped {result.Violations.Count} violations");

var cleanSvg = result.Svg;

CLI

dotnet tool install -g Mermaider.Cli

echo 'graph TD
  A --> B' | mermaid > diagram.svg

mermaid input.mmd -o output.svg --theme github-dark
mermaid --list-themes

MSAGL Layout Provider

If you prefer MSAGL for its edge routing fidelity on complex graphs, install the optional package:

dotnet add package Mermaider.Layout.Msagl
using Mermaider.Layout.Msagl;

// Global — all subsequent renders use MSAGL:
MermaidRenderer.SetLayoutProvider(new MsaglLayoutProvider());

// Or per-call:
var svg = MermaidRenderer.RenderSvg(input, new RenderOptions
{
    LayoutProvider = new MsaglLayoutProvider(),
});

AOT Support

Mermaider is fully compatible with .NET Native AOT. To publish your own AOT app:

<PropertyGroup>
  <PublishAot>true</PublishAot>
</PropertyGroup>

<ItemGroup>
  <PackageReference Include="Mermaider" />
</ItemGroup>
dotnet publish -c Release

Benchmarks

Graph-based diagram types use the built-in Sugiyama engine. Measured with [MemoryDiagnoser] on .NET 10 (Apple M2 Pro):

Method Mean Allocated
Flowchart (simple) ~23 µs ~46 KB
Flowchart (large) ~71 µs ~145 KB
Sequence ~12 µs ~28 KB
State ~17 µs ~47 KB
Class ~13 µs ~36 KB
ER ~17 µs ~45 KB
dotnet run --project tests/Mermaider.Benchmarks -c Release

Building from Source

git clone https://github.com/nullean/mermaider.git
cd mermaider
./build.sh build
./build.sh test

Attribution

This project started as a .NET port of beautiful-mermaid by Craft Docs (lukilabs). Their TypeScript library pioneered the idea of rendering Mermaid diagrams without a browser or DOM—fast, themeable, and synchronous.

beautiful-mermaid itself credits mermaid-ascii by Alexander Grooff for its ASCII rendering engine, which was ported from Go to TypeScript and extended.

beautiful-mermaid, relies on an external battle hardened layout engine elk.js,

We owe a huge thank-you to both projects for the excellent foundation.

A note on how this was built

This codebase was written with a coding agent (Claude). That said, care was taken to follow modern .NET 10 idioms and keep allocations low: ReadOnlySpan<char> parsing, [GeneratedRegex] with ReDoS timeout guards, FrozenDictionary / FrozenSet for hot-path lookups, SearchValues<char> for character classification, object pooling, and file-scoped namespaces throughout. The benchmark numbers above reflect the result.

License

MIT — see LICENSE.txt.

About

A pure dotnet mermaid parser, layout engine AND renderer, no js runtime, AOT ready.

Resources

License

Stars

Watchers

Forks

Packages

 
 
 

Contributors