[OpenAPI] Allow for invalid documents

[tobias-tengler]

Previously invalid documents would cause an endpoint to not be added. This means that a consumer would suddenly get a 404 instead of a more appropriate 500 statuscode.
This PR also removes validation that spans all documents and simplifies the validator to only validate on a per document basis.

[Copilot]

Pull request overview

This PR modifies the OpenAPI adapter to allow invalid documents to still create endpoints that return HTTP 500 instead of HTTP 404. Previously, invalid documents would prevent endpoints from being added entirely. The PR simplifies the validation approach by:

• Removing cross-document validation rules (e.g., model name uniqueness, route uniqueness, model references existence)
• Converting validation from async to sync operations
• Adding runtime validation when endpoints are invoked to return HTTP 500 for invalid documents
• Handling duplicate routes, operation names, and model names by using "first-wins" semantics with TryAdd

Key Changes

• Simplified validation to per-document basis only, removing cross-document validation rules
• Endpoints with invalid documents are now created but return HTTP 500 when invoked
• Changed validation methods from async to sync throughout the validation infrastructure

Reviewed changes

Copilot reviewed 40 out of 40 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
ValidationTestBase.cs	Removed tests for cross-document validation rules that were deleted
OpenApiIntegrationTestBase.cs	Added new tests for handling invalid documents (missing fields, missing model references, duplicated routes/operations/models)
HttpEndpointIntegrationTestBase.cs	Added tests verifying invalid endpoints return HTTP 500; minor formatting change to dictionary initializer
OpenApiDefinitionValidator.cs	Simplified validator to only include per-document validation rules; changed from async to sync
OpenApiDefinitionRegistry.cs	Removed complex cross-document validation logic; added duplicate detection for routes; uses TryAdd for model lookup
OpenApiEndpointFactory.cs	Added runtime document validation and stores validation result in endpoint descriptor; handles missing model references gracefully
OpenApiEndpointDescriptor.cs	Added HasValidDocument flag to track validation status
DynamicEndpointMiddleware.cs	Added check to return HTTP 500 for invalid documents before executing requests
DynamicOpenApiDocumentTransformer.cs	Added error handling with try-catch blocks; handles duplicate operation IDs with counters; uses TryAdd for components
Multiple validation rule files	Deleted cross-document validation rules (model name uniqueness, route uniqueness, model references, schema compilation)
Multiple validation interface files	Changed from async to sync validation methods
Multiple snapshot files	Added expected OpenAPI outputs for invalid document scenarios

Comments suppressed due to low confidence (1)

src/HotChocolate/Adapters/src/Adapters.OpenApi.Core/OpenApiDefinitionRegistry.cs:166

• The catch block silently swallows all exceptions without logging or diagnostics. This makes it difficult to debug issues when endpoint construction fails. Consider logging the exception or emitting a diagnostic event to help developers understand why an endpoint failed to be created.

            catch
            {
                // If the construction of an endpoint fails, we just skip over it.
            }

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[github-actions]

🚀 Fusion Gateway Performance Results

Simple Composite Query

Constant Load (50 VUs)

Requests/sec	Error Rate
5837.58 req/s	0.00%

📊 Response Time Metrics

Min	Med	Max	Avg	P90	P95
0.79ms	7.09ms	160.41ms	8.41ms	12.99ms	18.27ms

Ramping Load (0→50→500→50 VUs)

Requests/sec	Error Rate
4602.75 req/s	0.00%

📊 Response Time Metrics

Min	Med	Max	Avg	P90	P95
0.85ms	37.77ms	264.83ms	48.26ms	108.73ms	125.59ms

Executed Query

fragment User on User {
  id
  username
  name
}

fragment Review on Review {
  id
  body
}

fragment Product on Product {
  inStock
  name
  price
  shippingEstimate
  upc
  weight
}

query TestQuery {
  topProducts(first: 5) {
    ...Product
    reviews {
      ...Review
      author {
        ...User
      }
    }
  }
}

Deep Recursion Query

Constant Load (50 VUs)

Requests/sec	Error Rate
266.94 req/s	0.00%

📊 Response Time Metrics

Min	Med	Max	Avg	P90	P95
12.97ms	175.17ms	609.36ms	181.60ms	227.25ms	255.62ms

Ramping Load (0→50→500→50 VUs)

Requests/sec	Error Rate
308.29 req/s	0.00%

📊 Response Time Metrics

Min	Med	Max	Avg	P90	P95
3.03ms	675.95ms	1728.61ms	687.86ms	1392.43ms	1495.89ms

Executed Query

fragment User on User {
  id
  username
  name
}

fragment Review on Review {
  id
  body
}

fragment Product on Product {
  inStock
  name
  price
  shippingEstimate
  upc
  weight
}

query TestQuery {
  users {
    ...User
    reviews {
      ...Review
      product {
        ...Product
        reviews {
          ...Review
          author {
            ...User
            reviews {
              ...Review
              product {
                ...Product
              }
            }
          }
        }
      }
    }
  }
  topProducts(first: 5) {
    ...Product
    reviews {
      ...Review
      author {
        ...User
        reviews {
          ...Review
          product {
            ...Product
          }
        }
      }
    }
  }
}

Variable Batching Throughput

Constant Load (50 VUs)

Requests/sec	Error Rate
24000.08 req/s	0.00%

📊 Response Time Metrics

Min	Med	Max	Avg	P90	P95
0.09ms	1.68ms	41.82ms	2.03ms	3.89ms	4.74ms

Ramping Load (0→50→500→50 VUs)

Requests/sec	Error Rate
18898.56 req/s	0.00%

📊 Response Time Metrics

Min	Med	Max	Avg	P90	P95
0.09ms	9.22ms	97.23ms	11.20ms	23.17ms	27.92ms

Executed Query

query TestQuery_8f7a46ce_2(
  $__fusion_1_upc: ID!
  $__fusion_2_price: Long!
  $__fusion_2_weight: Long!
) {
  productByUpc(upc: $__fusion_1_upc) {
    inStock
    shippingEstimate(weight: $__fusion_2_weight, price: $__fusion_2_price)
  }
}

Variables (5 sets batched in single request)

[
  { "__fusion_1_upc": "1", "__fusion_2_price": 899, "__fusion_2_weight": 100 },
  { "__fusion_1_upc": "2", "__fusion_2_price": 1299, "__fusion_2_weight": 1000 },
  { "__fusion_1_upc": "3", "__fusion_2_price": 15, "__fusion_2_weight": 20 },
  { "__fusion_1_upc": "4", "__fusion_2_price": 499, "__fusion_2_weight": 100 },
  { "__fusion_1_upc": "5", "__fusion_2_price": 1299, "__fusion_2_weight": 1000 }
]

No baseline data available for comparison.

---

Run 20575417663 • Commit a6e50fc • Mon, 29 Dec 2025 14:57:18 GMT

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 0.00%. Comparing base (3b85c32) to head (13b6113).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@     Coverage Diff      @@
##   main   #9022   +/-   ##
============================
============================

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.