Fix render named-argument shadowing by evaluating assignments before applying them

[Copilot]

render keyword arguments were applied sequentially in the active render scope, so an early assignment could overwrite a variable needed by later argument expressions (e.g. value: value.f1, key: value.f2). This produced corrupted results when argument names overlapped with source objects.

• Behavioral fix (RenderStatement)

  • Changed render named-argument handling to:
    1. evaluate all assignment expressions first,
    2. then apply all resulting values to the render scope.
  • This preserves the original lookup context across all argument evaluations and aligns with expected Liquid behavior.

• Regression coverage

  • Added a focused test for:
    • source: {% render 'file', value: value.f1, key: value.f2 %}
    • snippet: {{ value }} {{ key }}
    • expected output: Hello World

• Scope of change

  • Limited to render assignment evaluation flow; no parser or syntax changes.

{% render 'file', value: value.f1, key: value.f2 %}

Before: first assignment could replace value and break value.f2 resolution.
After: both expressions resolve against the same pre-assignment context, then are assigned.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>Bug: Variable corruption in render tag assignments</issue_title>
<issue_description>Found a bug while working with Shopify's Horizon theme. When a render tag has multiple assignments and one of the parameter names matches an object that's referenced in later assignments, the object value gets corrupted.

list-filter.liquid#L369-L378:

{% render 'checkbox',
  name: value.param_name,
  value: value.value,
  label: value.label,
  checked: value.active,
  id: input_id,
  disabled: is_disabled,
  inputRef: 'facetInputs[]',
  events: 'on:pointerenter="/prefetchPage" on:pointerleave="/cancelPrefetchPage"',
  autofocus: autofocus
%}

Notice how value is used as both a parameter name and as the object being accessed (value.value, value.label, etc.).

Sample

var source = "{% render 'file', value: value.f1, key: value.f2 %}";
var file = "{{value}} {{key}}";

if (new FluidParser().TryParse(source, out var template, out var error))
{
    var context = new TemplateContext();
	
    var fileProvider = new MockFileProvider();
    fileProvider.Add("file", file);
    context.Options.FileProvider = fileProvider;

    context.SetValue("value", new MyClass { f1 = "Hello", f2 = "World" });

    Console.WriteLine(template.Render(context));
}

Expected output:

Hello World

Actual output:

Hello 

Analysis

It appears that assignments are being evaluated and applied to the context sequentially, so when value: value.f1 is processed:

1. value.f1 is evaluated correctly as "Hello"
2. value is set to "Hello" in the context
3. When key: value.f2 is evaluated, value is now "Hello" (a string), not the original object, so .f2 fails

The assignments should all be evaluated first using the original context, then applied together.</issue_description>

Comments on the Issue (you are @copilot in this section)

@hishamco Please submit a PR, or I could have a look @sebastienros Can you try with [3.0.0-beta.4](https://www.nuget.org/packages/Fluid.Core/3.0.0-beta.4) which I just published? It contains tons of bug fixes to align with Shopify's implementation. This was done by working on all the tests provided by https://github.com/jg-rp/golden-liquid

And please report on the other issues you filed after testing with this version, where it applies.</comment_new>

• Fixes Bug: Variable corruption in render tag assignments #915

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.