pipeline-bridge/e2e_test.go

package main

import (
	"encoding/json"
	"net/http"
	"net/http/httptest"
	"sync/atomic"
	"testing"

	"git.daviestechlabs.io/daviestechlabs/handler-base/messages"
)

// ────────────────────────────────────────────────────────────────────────────
// E2E tests: Argo + Kubeflow pipeline submission integration
// ────────────────────────────────────────────────────────────────────────────

func TestSubmitArgoE2E_FullPayload(t *testing.T) {
	// Verify the full Argo workflow structure
	var receivedBody map[string]any

	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		_ = json.NewDecoder(r.Body).Decode(&receivedBody)
		_ = json.NewEncoder(w).Encode(map[string]any{
			"metadata": map[string]any{"name": "doc-ingest-xyz"},
		})
	}))
	defer ts.Close()

	ctx := t.Context()
	params := map[string]any{
		"source":     "s3://bucket/docs",
		"collection": "knowledge-base",
		"batch_size": 100,
	}

	runID, err := submitArgo(ctx, ts.Client(), ts.URL, "ai-ml", "document-ingestion", params, "req-e2e-001")
	if err != nil {
		t.Fatal(err)
	}
	if runID != "doc-ingest-xyz" {
		t.Errorf("runID = %q", runID)
	}

	// Verify workflow structure
	wf, ok := receivedBody["workflow"].(map[string]any)
	if !ok {
		t.Fatal("missing workflow key")
	}
	meta := wf["metadata"].(map[string]any)
	if meta["namespace"] != "ai-ml" {
		t.Errorf("namespace = %v", meta["namespace"])
	}
	labels := meta["labels"].(map[string]any)
	if labels["request-id"] != "req-e2e-001" {
		t.Errorf("request-id label = %v", labels["request-id"])
	}

	spec := wf["spec"].(map[string]any)
	templateRef := spec["workflowTemplateRef"].(map[string]any)
	if templateRef["name"] != "document-ingestion" {
		t.Errorf("template = %v", templateRef["name"])
	}

	args := spec["arguments"].(map[string]any)
	argoParams := args["parameters"].([]any)
	if len(argoParams) != 3 {
		t.Errorf("param count = %d, want 3", len(argoParams))
	}
}

func TestSubmitKubeflowE2E_FullPayload(t *testing.T) {
	var receivedBody map[string]any

	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		_ = json.NewDecoder(r.Body).Decode(&receivedBody)
		_ = json.NewEncoder(w).Encode(map[string]any{
			"run": map[string]any{"id": "kf-run-e2e-789"},
		})
	}))
	defer ts.Close()

	ctx := t.Context()
	params := map[string]any{
		"query":   "what is kubernetes",
		"top_k":   5,
		"user_id": "test-user",
	}

	runID, err := submitKubeflow(ctx, ts.Client(), ts.URL, "rag-pipeline", params, "req-e2e-002")
	if err != nil {
		t.Fatal(err)
	}
	if runID != "kf-run-e2e-789" {
		t.Errorf("runID = %q", runID)
	}

	// Verify run name format
	name := receivedBody["name"].(string)
	if name != "rag-pipeline-req-e2e-" {
		t.Errorf("run name = %q", name)
	}
}

func TestPipelineDispatchE2E_AllEngines(t *testing.T) {
	// Verify all pipeline definitions dispatch to correct engine
	for name, def := range pipelines {
		t.Run(name, func(t *testing.T) {
			if def.Engine != "argo" && def.Engine != "kubeflow" {
				t.Errorf("engine = %q, want argo or kubeflow", def.Engine)
			}
			if def.Engine == "argo" && def.Template == "" {
				t.Errorf("argo pipeline %q missing template", name)
			}
			if def.Engine == "kubeflow" && def.PipelineID == "" {
				t.Errorf("kubeflow pipeline %q missing pipeline_id", name)
			}
		})
	}
}

func TestPipelineDispatchE2E_UnknownPipeline(t *testing.T) {
	// Verify unknown pipeline is rejected and available list is provided
	pipelineName := "nonexistent-pipeline"
	_, ok := pipelines[pipelineName]
	if ok {
		t.Error("nonexistent pipeline should not be found")
	}

	names := make([]string, 0, len(pipelines))
	for k := range pipelines {
		names = append(names, k)
	}
	resp := &messages.PipelineStatus{
		RequestID:          "req-bad",
		Status:             "error",
		Error:              "Unknown pipeline: nonexistent-pipeline",
		AvailablePipelines: names,
	}

	if resp.Status != "error" {
		t.Error("expected error status")
	}
	if len(resp.AvailablePipelines) != len(pipelines) {
		t.Errorf("available_pipelines count mismatch")
	}
}

func TestSubmitArgoE2E_ConcurrentRequests(t *testing.T) {
	var count atomic.Int64

	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		count.Add(1)
		_ = json.NewEncoder(w).Encode(map[string]any{
			"metadata": map[string]any{"name": "concurrent-wf"},
		})
	}))
	defer ts.Close()

	ctx := t.Context()
	errs := make(chan error, 10)

	for i := 0; i < 10; i++ {
		go func() {
			_, err := submitArgo(ctx, ts.Client(), ts.URL, "ai-ml", "batch-inference",
				map[string]any{"batch": i}, "req-concurrent")
			errs <- err
		}()
	}

	for i := 0; i < 10; i++ {
		if err := <-errs; err != nil {
			t.Errorf("concurrent request %d: %v", i, err)
		}
	}

	if got := count.Load(); got != 10 {
		t.Errorf("request count = %d, want 10", got)
	}
}

// ────────────────────────────────────────────────────────────────────────────
// Benchmarks
// ────────────────────────────────────────────────────────────────────────────

func BenchmarkSubmitArgo(b *testing.B) {
	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		_, _ = w.Write([]byte(`{"metadata":{"name":"bench-wf"}}`))
	}))
	defer ts.Close()

	ctx := b.Context()
	params := map[string]any{"source": "test"}

	b.ResetTimer()
	for b.Loop() {
		_, _ = submitArgo(ctx, ts.Client(), ts.URL, "ai-ml", "document-ingestion", params, "bench-req")
	}
}

func BenchmarkSubmitKubeflow(b *testing.B) {
	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		_, _ = w.Write([]byte(`{"run":{"id":"bench-run"}}`))
	}))
	defer ts.Close()

	ctx := b.Context()
	params := map[string]any{"query": "test"}

	b.ResetTimer()
	for b.Loop() {
		_, _ = submitKubeflow(ctx, ts.Client(), ts.URL, "rag-pipeline", params, "bench-req")
	}
}