yavsc/src/Yavsc.Blogs.Tests/BlogApiTests.cs

245 lines
9.7 KiB
C#
Raw Normal View History

using System.Net;
using System.Net.Http;
using System.Net.Http.Json;
using System.Text.Json;
using Microsoft.Extensions.DependencyInjection;
using Yavsc.Models;
using Yavsc.Models.Blog;
using Yavsc.Tests.Shared;
namespace Yavsc.Blogs.Tests;
/// <summary>
/// Behavioural tests for <c>BlogApiController</c>. Built on the
/// <see cref="BlogsWebServerFixture"/> scaffold: in-memory
test(blogs): real JwtBearer in fixture, drop X-Test-Role bypass Wire the Blogs integration test host with a real AddJwtBearer (HS256, IssuerSigningKey shared with the new TestTokenIssuer) and the production BlogScope policy verbatim, instead of the TestAuthPolicyProvider / AllowAllAuthorizationService / NoopAuthHandler stack that short-circuited every authorization check. Why: BlogSpotService.Modify calls IAuthorizationService.AuthorizeAsync(user, blog, EditPermission); the previous AllowAllAuthorizationService stub made that a no-op, so the tests could not exercise the real ownership chain and any change in PermissionHandler would silently slip through. The new test host registers the real PermissionHandler, so a PUT that succeeds (204) is now proof that PermissionHandler.IsOwner accepted the request — i.e. the JWT's sub matched the post's AuthorId, end-to-end. Notes for future-me: - JwtSecurityTokenHandler.DefaultInboundClaimTypeMap.Clear() is called once on the first Issue() to keep the 'sub' claim literal; without it UserHelpers.GetUserId (which reads 'sub') gets ClaimTypes.NameIdentifier instead, returns null, and the owner check fails for every PUT. The companion options.MapInboundClaims = false on the validation pipeline keeps both sides in sync. - Production still uses AddYavscJwtBearer against the OIDC authority; the test-only HS256 path is local to the test process and never crosses a network boundary. Coverage: - GetBlog_returns_401_when_no_token_is_provided — anonymous request, real policy fails closed. - PutBlog_with_valid_token_and_owner_returns_204_and_Get_ reflects_update — POST then PUT then GET, all behind a real JWT, asserting 204 + list contains the updated title. Packages added to Directory.Packages.props at 8.2.1 to match what Microsoft.AspNetCore.Authentication.JwtBearer 10.0.9 already transitively pulls in (no version drift).
2026-07-06 23:29:51 +01:00
/// <c>ApplicationDbContext</c>, real <c>BlogSpotService</c>, and a
/// real <c>AddJwtBearer</c> validating HS256 tokens signed by
/// <see cref="TestTokenIssuer"/>. The production <c>BlogScope</c>
/// policy runs unmodified — sending <c>Authorization: Bearer …</c>
/// with a valid token is what gets a request through, omitting the
/// header (or sending a token signed with the wrong key) gets a
/// 401 back from the framework.
/// </summary>
public sealed class BlogApiTests : IClassFixture<BlogsWebServerFixture>
{
private readonly BlogsWebServerFixture _fixture;
public BlogApiTests(BlogsWebServerFixture fixture)
{
_fixture = fixture;
}
/// <summary>Reset the in-memory database to a known empty state.
/// <c>UseInMemoryDatabase</c> shares its store across the
/// lifetime of the <see cref="BlogsWebServerFixture"/> instance,
/// so without a per-test reset the test order would leak
/// state between tests.</summary>
private void ResetDatabase()
{
using var scope = _fixture.Services.CreateScope();
var db = scope.ServiceProvider.GetRequiredService<ApplicationDbContext>();
db.Database.EnsureDeleted();
db.Database.EnsureCreated();
}
/// <summary>The fixture's <c>WebApplication</c> is bound to
/// <c>https://localhost:&lt;random&gt;</c> via
/// <see cref="WebHostFixture.Addresses"/>. We pick the first
/// https URL and append the controller route
/// (<c>/api/v1/blog</c>, matching the production
/// <c>[Route(APIPrefix + "/blog")]</c>).</summary>
private string BlogsUrl =>
_fixture.Addresses.First(a => a.StartsWith("https://")) + "/api/v1/blog";
test(blogs): real JwtBearer in fixture, drop X-Test-Role bypass Wire the Blogs integration test host with a real AddJwtBearer (HS256, IssuerSigningKey shared with the new TestTokenIssuer) and the production BlogScope policy verbatim, instead of the TestAuthPolicyProvider / AllowAllAuthorizationService / NoopAuthHandler stack that short-circuited every authorization check. Why: BlogSpotService.Modify calls IAuthorizationService.AuthorizeAsync(user, blog, EditPermission); the previous AllowAllAuthorizationService stub made that a no-op, so the tests could not exercise the real ownership chain and any change in PermissionHandler would silently slip through. The new test host registers the real PermissionHandler, so a PUT that succeeds (204) is now proof that PermissionHandler.IsOwner accepted the request — i.e. the JWT's sub matched the post's AuthorId, end-to-end. Notes for future-me: - JwtSecurityTokenHandler.DefaultInboundClaimTypeMap.Clear() is called once on the first Issue() to keep the 'sub' claim literal; without it UserHelpers.GetUserId (which reads 'sub') gets ClaimTypes.NameIdentifier instead, returns null, and the owner check fails for every PUT. The companion options.MapInboundClaims = false on the validation pipeline keeps both sides in sync. - Production still uses AddYavscJwtBearer against the OIDC authority; the test-only HS256 path is local to the test process and never crosses a network boundary. Coverage: - GetBlog_returns_401_when_no_token_is_provided — anonymous request, real policy fails closed. - PutBlog_with_valid_token_and_owner_returns_204_and_Get_ reflects_update — POST then PUT then GET, all behind a real JWT, asserting 204 + list contains the updated title. Packages added to Directory.Packages.props at 8.2.1 to match what Microsoft.AspNetCore.Authentication.JwtBearer 10.0.9 already transitively pulls in (no version drift).
2026-07-06 23:29:51 +01:00
/// <summary>Build an authenticated client: a real
/// <c>Authorization: Bearer &lt;jwt&gt;</c> header where the JWT
/// is signed by <see cref="TestTokenIssuer"/> and carries
/// <c>sub = subject</c>. The production <c>BlogScope</c> policy
/// reads <c>scope=blogs</c> off the same token, so
/// <c>TestTokenIssuer.Issue</c>'s default scope is enough.</summary>
private HttpClient NewClient(string subject = "tester")
{
// The fixture's self-signed certificate is not in the user's
// trust store, so we accept anything (same pattern as
// Yavsc.Org.Tests' BypassSslValidationHandler).
var handler = new HttpClientHandler
{
ServerCertificateCustomValidationCallback = (_, _, _, _) => true
};
var http = new HttpClient(handler)
{
BaseAddress = new Uri(_fixture.Addresses.First(a => a.StartsWith("https://")))
};
test(blogs): real JwtBearer in fixture, drop X-Test-Role bypass Wire the Blogs integration test host with a real AddJwtBearer (HS256, IssuerSigningKey shared with the new TestTokenIssuer) and the production BlogScope policy verbatim, instead of the TestAuthPolicyProvider / AllowAllAuthorizationService / NoopAuthHandler stack that short-circuited every authorization check. Why: BlogSpotService.Modify calls IAuthorizationService.AuthorizeAsync(user, blog, EditPermission); the previous AllowAllAuthorizationService stub made that a no-op, so the tests could not exercise the real ownership chain and any change in PermissionHandler would silently slip through. The new test host registers the real PermissionHandler, so a PUT that succeeds (204) is now proof that PermissionHandler.IsOwner accepted the request — i.e. the JWT's sub matched the post's AuthorId, end-to-end. Notes for future-me: - JwtSecurityTokenHandler.DefaultInboundClaimTypeMap.Clear() is called once on the first Issue() to keep the 'sub' claim literal; without it UserHelpers.GetUserId (which reads 'sub') gets ClaimTypes.NameIdentifier instead, returns null, and the owner check fails for every PUT. The companion options.MapInboundClaims = false on the validation pipeline keeps both sides in sync. - Production still uses AddYavscJwtBearer against the OIDC authority; the test-only HS256 path is local to the test process and never crosses a network boundary. Coverage: - GetBlog_returns_401_when_no_token_is_provided — anonymous request, real policy fails closed. - PutBlog_with_valid_token_and_owner_returns_204_and_Get_ reflects_update — POST then PUT then GET, all behind a real JWT, asserting 204 + list contains the updated title. Packages added to Directory.Packages.props at 8.2.1 to match what Microsoft.AspNetCore.Authentication.JwtBearer 10.0.9 already transitively pulls in (no version drift).
2026-07-06 23:29:51 +01:00
http.DefaultRequestHeaders.Authorization =
new System.Net.Http.Headers.AuthenticationHeaderValue(
"Bearer", TestTokenIssuer.Issue(subject));
return http;
}
test(blogs): real JwtBearer in fixture, drop X-Test-Role bypass Wire the Blogs integration test host with a real AddJwtBearer (HS256, IssuerSigningKey shared with the new TestTokenIssuer) and the production BlogScope policy verbatim, instead of the TestAuthPolicyProvider / AllowAllAuthorizationService / NoopAuthHandler stack that short-circuited every authorization check. Why: BlogSpotService.Modify calls IAuthorizationService.AuthorizeAsync(user, blog, EditPermission); the previous AllowAllAuthorizationService stub made that a no-op, so the tests could not exercise the real ownership chain and any change in PermissionHandler would silently slip through. The new test host registers the real PermissionHandler, so a PUT that succeeds (204) is now proof that PermissionHandler.IsOwner accepted the request — i.e. the JWT's sub matched the post's AuthorId, end-to-end. Notes for future-me: - JwtSecurityTokenHandler.DefaultInboundClaimTypeMap.Clear() is called once on the first Issue() to keep the 'sub' claim literal; without it UserHelpers.GetUserId (which reads 'sub') gets ClaimTypes.NameIdentifier instead, returns null, and the owner check fails for every PUT. The companion options.MapInboundClaims = false on the validation pipeline keeps both sides in sync. - Production still uses AddYavscJwtBearer against the OIDC authority; the test-only HS256 path is local to the test process and never crosses a network boundary. Coverage: - GetBlog_returns_401_when_no_token_is_provided — anonymous request, real policy fails closed. - PutBlog_with_valid_token_and_owner_returns_204_and_Get_ reflects_update — POST then PUT then GET, all behind a real JWT, asserting 204 + list contains the updated title. Packages added to Directory.Packages.props at 8.2.1 to match what Microsoft.AspNetCore.Authentication.JwtBearer 10.0.9 already transitively pulls in (no version drift).
2026-07-06 23:29:51 +01:00
/// <summary>Build an unauthenticated client. Used to assert that
/// the <c>BlogScope</c> policy fails closed when no bearer
/// token is presented.</summary>
private HttpClient NewAnonymousClient()
{
var handler = new HttpClientHandler
{
ServerCertificateCustomValidationCallback = (_, _, _, _) => true
};
return new HttpClient(handler)
{
BaseAddress = new Uri(_fixture.Addresses.First(a => a.StartsWith("https://")))
};
}
[Fact]
public async Task GetBlogs_returns_200_with_empty_list_when_no_posts()
{
ResetDatabase();
using var http = NewClient();
var response = await http.GetAsync("/api/v1/blog");
Assert.Equal(HttpStatusCode.OK, response.StatusCode);
var body = await response.Content.ReadAsStringAsync();
// Empty table → empty JSON array. We compare as a JsonDocument
// so a future change in formatting (whitespace, indentation)
// doesn't break the assertion.
using var doc = JsonDocument.Parse(body);
Assert.Equal(JsonValueKind.Array, doc.RootElement.ValueKind);
Assert.Equal(0, doc.RootElement.GetArrayLength());
}
[Fact]
public async Task PostBlog_creates_a_post_and_Get_returns_it_in_the_list()
{
ResetDatabase();
using var http = NewClient();
// Create a minimal BlogPost. The server assigns Id, so we
// send 0 + an explicit AuthorId; the production
// BlogSpotService.Create() tolerates that.
var draft = new BlogPost
{
Id = 0,
Title = "Premier billet",
AuthorId = "tester",
Article = "Contenu de test.",
DateCreated = DateTime.UtcNow,
DateModified = DateTime.UtcNow
};
var postResponse = await http.PostAsJsonAsync("/api/v1/blog", draft);
Assert.Equal(HttpStatusCode.Created, postResponse.StatusCode);
// The POST returns the server-issued post (with a real Id).
var created = await postResponse.Content.ReadFromJsonAsync<BlogPost>();
Assert.NotNull(created);
Assert.NotEqual(0, created!.Id);
Assert.Equal(draft.Title, created.Title);
// The list should now contain exactly one entry.
var listResponse = await http.GetAsync("/api/v1/blog");
Assert.Equal(HttpStatusCode.OK, listResponse.StatusCode);
using var doc = JsonDocument.Parse(await listResponse.Content.ReadAsStringAsync());
Assert.Equal(JsonValueKind.Array, doc.RootElement.ValueKind);
Assert.Equal(1, doc.RootElement.GetArrayLength());
Assert.Equal(created.Id, doc.RootElement[0].GetProperty("id").GetInt64());
}
test(blogs): real JwtBearer in fixture, drop X-Test-Role bypass Wire the Blogs integration test host with a real AddJwtBearer (HS256, IssuerSigningKey shared with the new TestTokenIssuer) and the production BlogScope policy verbatim, instead of the TestAuthPolicyProvider / AllowAllAuthorizationService / NoopAuthHandler stack that short-circuited every authorization check. Why: BlogSpotService.Modify calls IAuthorizationService.AuthorizeAsync(user, blog, EditPermission); the previous AllowAllAuthorizationService stub made that a no-op, so the tests could not exercise the real ownership chain and any change in PermissionHandler would silently slip through. The new test host registers the real PermissionHandler, so a PUT that succeeds (204) is now proof that PermissionHandler.IsOwner accepted the request — i.e. the JWT's sub matched the post's AuthorId, end-to-end. Notes for future-me: - JwtSecurityTokenHandler.DefaultInboundClaimTypeMap.Clear() is called once on the first Issue() to keep the 'sub' claim literal; without it UserHelpers.GetUserId (which reads 'sub') gets ClaimTypes.NameIdentifier instead, returns null, and the owner check fails for every PUT. The companion options.MapInboundClaims = false on the validation pipeline keeps both sides in sync. - Production still uses AddYavscJwtBearer against the OIDC authority; the test-only HS256 path is local to the test process and never crosses a network boundary. Coverage: - GetBlog_returns_401_when_no_token_is_provided — anonymous request, real policy fails closed. - PutBlog_with_valid_token_and_owner_returns_204_and_Get_ reflects_update — POST then PUT then GET, all behind a real JWT, asserting 204 + list contains the updated title. Packages added to Directory.Packages.props at 8.2.1 to match what Microsoft.AspNetCore.Authentication.JwtBearer 10.0.9 already transitively pulls in (no version drift).
2026-07-06 23:29:51 +01:00
[Fact]
public async Task GetBlog_returns_401_when_no_token_is_provided()
{
ResetDatabase();
using var http = NewAnonymousClient();
// No Authorization header → the JwtBearer middleware
// produces an unauthenticated principal, the BlogScope
// policy's RequireAuthenticatedUser requirement fails, and
// the framework returns 401. This is the proof that the
// production policy is wired in the test host and not
// short-circuited by a test-only auth bypass.
var response = await http.GetAsync("/api/v1/blog");
Assert.Equal(HttpStatusCode.Unauthorized, response.StatusCode);
}
[Fact]
public async Task PutBlog_with_valid_token_and_owner_returns_204_and_Get_reflects_update()
{
ResetDatabase();
// The JWT's sub must match the post's AuthorId:
// PermissionHandler.IsOwner checks blog.AuthorId == user.GetUserId(),
// and UserHelpers.GetUserId reads "sub" off the principal.
// A mismatched sub → AuthorizationFailureException →
// Challenge() (401) from the controller. The 204 in this
// test is the proof that the real authorization chain
// accepted the request, end-to-end.
using var http = NewClient(subject: "tester");
// Seed a post we can update.
var draft = new BlogPost
{
Id = 0,
Title = "Avant",
AuthorId = "tester",
Article = "Contenu initial.",
DateCreated = DateTime.UtcNow,
DateModified = DateTime.UtcNow
};
var postResponse = await http.PostAsJsonAsync("/api/v1/blog", draft);
Assert.Equal(HttpStatusCode.Created, postResponse.StatusCode);
var created = (await postResponse.Content.ReadFromJsonAsync<BlogPost>())!;
// PUT with the server-issued Id; the controller rejects
// mismatched id/blog.Id with 400, so we keep them aligned.
var update = new BlogPost
{
Id = created.Id,
Title = "Après",
AuthorId = created.AuthorId,
Article = created.Article,
DateCreated = created.DateCreated,
DateModified = DateTime.UtcNow
};
var putResponse = await http.PutAsJsonAsync($"/api/v1/blog/{created.Id}", update);
Assert.Equal(HttpStatusCode.NoContent, putResponse.StatusCode);
// The list should now reflect the new title.
var listResponse = await http.GetAsync("/api/v1/blog");
Assert.Equal(HttpStatusCode.OK, listResponse.StatusCode);
using var doc = JsonDocument.Parse(await listResponse.Content.ReadAsStringAsync());
Assert.Equal(JsonValueKind.Array, doc.RootElement.ValueKind);
Assert.Equal(1, doc.RootElement.GetArrayLength());
Assert.Equal("Après", doc.RootElement[0].GetProperty("title").GetString());
}
[Fact]
public async Task DeleteBlog_removes_a_post_and_Get_returns_an_empty_list()
{
ResetDatabase();
using var http = NewClient();
// Seed a post we can delete.
var draft = new BlogPost
{
Id = 0,
Title = "À supprimer",
AuthorId = "tester",
Article = "Contenu.",
DateCreated = DateTime.UtcNow,
DateModified = DateTime.UtcNow
};
var postResponse = await http.PostAsJsonAsync("/api/v1/blog", draft);
var created = (await postResponse.Content.ReadFromJsonAsync<BlogPost>())!;
var deleteResponse = await http.DeleteAsync($"/api/v1/blog/{created.Id}");
Assert.Equal(HttpStatusCode.OK, deleteResponse.StatusCode);
// The list should now be empty.
var listResponse = await http.GetAsync("/api/v1/blog");
using var doc = JsonDocument.Parse(await listResponse.Content.ReadAsStringAsync());
Assert.Equal(0, doc.RootElement.GetArrayLength());
}
}