package ollama

import (
	"context"
	"encoding/json"
	"net/http"
	"net/http/httptest"
	"strings"
	"sync"
	"testing"

	"go.uber.org/zap"

	noderuntime "iop/apps/node/internal/runtime"
	"iop/packages/go/config"
)

type fakeSink struct {
	mu     sync.Mutex
	events []noderuntime.RuntimeEvent
}

func (s *fakeSink) Emit(_ context.Context, event noderuntime.RuntimeEvent) error {
	s.mu.Lock()
	defer s.mu.Unlock()
	s.events = append(s.events, event)
	return nil
}

func (s *fakeSink) all() []noderuntime.RuntimeEvent {
	s.mu.Lock()
	defer s.mu.Unlock()
	return append([]noderuntime.RuntimeEvent(nil), s.events...)
}

func TestOllamaExecuteStreamsChatDeltas(t *testing.T) {
	var gotModel string
	var gotPrompt string
	var gotNumCtx int
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if r.URL.Path != "/api/chat" {
			t.Fatalf("unexpected path %s", r.URL.Path)
		}
		var req ollamaChatRequest
		if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
			t.Fatalf("decode request: %v", err)
		}
		gotModel = req.Model
		if req.Options != nil {
			switch v := req.Options["num_ctx"].(type) {
			case float64:
				gotNumCtx = int(v)
			case int:
				gotNumCtx = v
			}
		}
		if len(req.Messages) != 1 {
			t.Fatalf("expected one message, got %+v", req.Messages)
		}
		gotPrompt = req.Messages[0].Content
		w.Header().Set("Content-Type", "application/x-ndjson")
		_, _ = w.Write([]byte(`{"message":{"role":"assistant","content":"hello "},"done":false}` + "\n"))
		_, _ = w.Write([]byte(`{"message":{"role":"assistant","content":"world"},"done":false}` + "\n"))
		_, _ = w.Write([]byte(`{"done":true,"prompt_eval_count":3,"eval_count":2}` + "\n"))
	}))
	defer server.Close()

	adapter := New(config.OllamaConf{BaseURL: server.URL, ContextSize: 262144}, zap.NewNop())
	sink := &fakeSink{}
	err := adapter.Execute(context.Background(), noderuntime.ExecutionSpec{
		RunID:  "run-1",
		Target: "llama-test",
		Input:  map[string]any{"prompt": "say hello"},
	}, sink)
	if err != nil {
		t.Fatalf("Execute failed: %v", err)
	}
	if gotModel != "llama-test" {
		t.Fatalf("model: got %q", gotModel)
	}
	if gotPrompt != "say hello" {
		t.Fatalf("prompt: got %q", gotPrompt)
	}
	if gotNumCtx != 262144 {
		t.Fatalf("num_ctx: got %d", gotNumCtx)
	}

	events := sink.all()
	if len(events) != 4 {
		t.Fatalf("expected 4 events, got %+v", events)
	}
	if events[0].Type != noderuntime.EventTypeStart {
		t.Fatalf("expected start event, got %+v", events[0])
	}
	if events[1].Delta+events[2].Delta != "hello world" {
		t.Fatalf("unexpected delta text: %+v", events)
	}
	if events[3].Type != noderuntime.EventTypeComplete {
		t.Fatalf("expected complete event, got %+v", events[3])
	}
	if events[3].Usage == nil || events[3].Usage.InputTokens != 3 || events[3].Usage.OutputTokens != 2 {
		t.Fatalf("unexpected usage: %+v", events[3].Usage)
	}
}

func TestOllamaExecutePassesOptionsAndTopLevelFields(t *testing.T) {
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if r.URL.Path != "/api/chat" {
			t.Fatalf("unexpected path %s", r.URL.Path)
		}
		var req ollamaChatRequest
		if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
			t.Fatalf("decode request: %v", err)
		}
		if req.Options["temperature"].(float64) != 0.2 || req.Options["top_p"].(float64) != 0.9 || req.Options["num_predict"].(float64) != 32 {
			t.Fatalf("unexpected options: %+v", req.Options)
		}
		switch v := req.Options["num_ctx"].(type) {
		case float64:
			if int(v) != 262144 {
				t.Fatalf("profile context_size should override request num_ctx: got %d, want 262144", int(v))
			}
		case int:
			if v != 262144 {
				t.Fatalf("profile context_size should override request num_ctx: got %d, want 262144", v)
			}
		default:
			t.Fatalf("unexpected num_ctx type %T in %+v", v, req.Options)
		}
		if req.KeepAlive != "10m" || req.Think != false || req.Format != "json" {
			t.Fatalf("top-level fields not passed: keep_alive=%v think=%v format=%v", req.KeepAlive, req.Think, req.Format)
		}
		if tools, ok := req.Tools.([]any); !ok || len(tools) != 1 {
			t.Fatalf("tools not passed: %+v", req.Tools)
		}
		w.Header().Set("Content-Type", "application/x-ndjson")
		_, _ = w.Write([]byte(`{"message":{"role":"assistant","content":"ok"},"done":false}` + "\n"))
		_, _ = w.Write([]byte(`{"done":true}` + "\n"))
	}))
	defer server.Close()

	adapter := New(config.OllamaConf{BaseURL: server.URL, ContextSize: 262144}, zap.NewNop())
	sink := &fakeSink{}
	err := adapter.Execute(context.Background(), noderuntime.ExecutionSpec{
		RunID:  "run-options",
		Target: "llama-test",
		Input: map[string]any{
			"prompt": "hi",
			"options": map[string]any{
				"temperature": 0.2,
				"top_p":       0.9,
				"num_predict": float64(32),
				"num_ctx":     float64(4096),
			},
			"keep_alive": "10m",
			"think":      false,
			"format":     "json",
			"tools":      []any{map[string]any{"type": "function"}},
		},
	}, sink)
	if err != nil {
		t.Fatalf("Execute failed: %v", err)
	}
}

func TestOllamaExecutePreservesNativeToolCalls(t *testing.T) {
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if r.URL.Path != "/api/chat" {
			t.Fatalf("unexpected path %s", r.URL.Path)
		}
		w.Header().Set("Content-Type", "application/x-ndjson")
		_, _ = w.Write([]byte(`{"message":{"role":"assistant","tool_calls":[{"id":"call_1","type":"function","function":{"name":"run_commands","arguments":"{\"commands\":[\"git status\"]}"}}]},"done":false}` + "\n"))
		_, _ = w.Write([]byte(`{"done":true}` + "\n"))
	}))
	defer server.Close()

	adapter := New(config.OllamaConf{BaseURL: server.URL}, zap.NewNop())
	sink := &fakeSink{}
	err := adapter.Execute(context.Background(), noderuntime.ExecutionSpec{
		RunID:  "run-tools",
		Target: "llama-test",
		Input:  map[string]any{"prompt": "status"},
	}, sink)
	if err != nil {
		t.Fatalf("Execute failed: %v", err)
	}
	events := sink.all()
	complete := events[len(events)-1]
	if complete.Metadata["finish_reason"] != "tool_calls" {
		t.Fatalf("finish_reason: %+v", complete.Metadata)
	}
	var toolCalls []map[string]any
	if err := json.Unmarshal([]byte(complete.Metadata[runtimeMetadataOpenAIToolCalls]), &toolCalls); err != nil {
		t.Fatalf("tool_calls metadata JSON: %v", err)
	}
	if len(toolCalls) != 1 || toolCalls[0]["id"] != "call_1" {
		t.Fatalf("tool_calls: %+v", toolCalls)
	}
}

func TestOllamaExecuteEmitsErrorForHTTPFailure(t *testing.T) {
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
		http.Error(w, "boom", http.StatusBadGateway)
	}))
	defer server.Close()

	adapter := New(config.OllamaConf{BaseURL: server.URL}, zap.NewNop())
	sink := &fakeSink{}
	err := adapter.Execute(context.Background(), noderuntime.ExecutionSpec{
		RunID:  "run-err",
		Target: "llama-test",
		Input:  map[string]any{"prompt": "hi"},
	}, sink)
	if err == nil {
		t.Fatal("expected error")
	}
	events := sink.all()
	if len(events) < 2 || events[1].Type != noderuntime.EventTypeError {
		t.Fatalf("expected error event, got %+v", events)
	}
	if !strings.Contains(events[1].Error, "boom") {
		t.Fatalf("expected HTTP body in error, got %q", events[1].Error)
	}
}

func TestOllamaCapabilitiesQueryTags(t *testing.T) {
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if r.URL.Path != "/api/tags" {
			t.Fatalf("unexpected path %s", r.URL.Path)
		}
		_, _ = w.Write([]byte(`{"models":[{"name":"llama-a"},{"name":"llama-b"}]}`))
	}))
	defer server.Close()

	adapter := New(config.OllamaConf{
		BaseURL:          server.URL,
		Capacity:         5,
		MaxQueue:         10,
		QueueTimeoutMS:   15000,
		RequestTimeoutMS: 30000,
	}, zap.NewNop())
	caps, err := adapter.Capabilities(context.Background())
	if err != nil {
		t.Fatalf("Capabilities failed: %v", err)
	}
	got := strings.Join(caps.Targets, ",")
	if got != "llama-a,llama-b" {
		t.Fatalf("targets: got %q", got)
	}
	if caps.MaxConcurrency != 5 {
		t.Fatalf("expected MaxConcurrency 5, got %d", caps.MaxConcurrency)
	}
	if caps.MaxQueue != 10 {
		t.Fatalf("expected MaxQueue 10, got %d", caps.MaxQueue)
	}
	if caps.QueueTimeoutMS != 15000 {
		t.Fatalf("expected QueueTimeoutMS 15000, got %d", caps.QueueTimeoutMS)
	}
	if caps.RequestTimeoutMS != 30000 {
		t.Fatalf("expected RequestTimeoutMS 30000, got %d", caps.RequestTimeoutMS)
	}

	// Verify default capacity fallback
	adapterDefault := New(config.OllamaConf{BaseURL: server.URL}, zap.NewNop())
	capsDefault, err := adapterDefault.Capabilities(context.Background())
	if err != nil {
		t.Fatalf("Capabilities failed: %v", err)
	}
	if capsDefault.MaxConcurrency != 4 {
		t.Fatalf("expected default MaxConcurrency 4, got %d", capsDefault.MaxConcurrency)
	}
}

func TestOllamaHandleCommandPassesNativeAPI(t *testing.T) {
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if r.Method != http.MethodPost || r.URL.RequestURI() != "/api/show?verbose=true" {
			t.Fatalf("unexpected request %s %s", r.Method, r.URL.RequestURI())
		}
		var body map[string]any
		if err := json.NewDecoder(r.Body).Decode(&body); err != nil {
			t.Fatalf("decode body: %v", err)
		}
		if body["model"] != "gemma4:26b" {
			t.Fatalf("body not passed: %+v", body)
		}
		w.Header().Set("Content-Type", "application/json")
		w.WriteHeader(http.StatusAccepted)
		_, _ = w.Write([]byte(`{"license":"ok"}`))
	}))
	defer server.Close()

	adapter := New(config.OllamaConf{BaseURL: server.URL}, zap.NewNop())
	resp, err := adapter.HandleCommand(context.Background(), noderuntime.CommandRequest{
		RequestID: "ollama-1",
		Type:      noderuntime.CommandTypeOllamaAPI,
		Adapter:   "ollama",
		Metadata: map[string]string{
			"ollama_method": "POST",
			"ollama_path":   "/api/show?verbose=true",
			"ollama_body":   `{"model":"gemma4:26b"}`,
		},
	})
	if err != nil {
		t.Fatalf("HandleCommand failed: %v", err)
	}
	if resp.Result["status_code"] != "202" || resp.Result["content_type"] != "application/json" || resp.Result["body"] != `{"license":"ok"}` {
		t.Fatalf("unexpected response: %+v", resp.Result)
	}
}

func TestOllamaMultiEndpoint(t *testing.T) {
	var (
		gotModelA  string
		gotNumCtxA int
		gotModelB  string
		gotNumCtxB int
	)

	makeServer := func(gotModel *string, gotNumCtx *int, reply string) *httptest.Server {
		return httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
			if r.URL.Path == "/api/tags" {
				_, _ = w.Write([]byte(`{"models":[]}`))
				return
			}
			if r.URL.Path != "/api/chat" {
				t.Errorf("unexpected path %s", r.URL.Path)
				return
			}
			var req ollamaChatRequest
			if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
				t.Errorf("decode request: %v", err)
				return
			}
			*gotModel = req.Model
			if v, ok := req.Options["num_ctx"].(float64); ok {
				*gotNumCtx = int(v)
			}
			w.Header().Set("Content-Type", "application/x-ndjson")
			_, _ = w.Write([]byte(`{"message":{"role":"assistant","content":"` + reply + `"},"done":false}` + "\n"))
			_, _ = w.Write([]byte(`{"done":true}` + "\n"))
		}))
	}

	serverA := makeServer(&gotModelA, &gotNumCtxA, "a")
	defer serverA.Close()
	serverB := makeServer(&gotModelB, &gotNumCtxB, "b")
	defer serverB.Close()

	adapterA := New(config.OllamaConf{BaseURL: serverA.URL, ContextSize: 4096}, zap.NewNop(), "ollama-a")
	adapterB := New(config.OllamaConf{BaseURL: serverB.URL, ContextSize: 8192}, zap.NewNop(), "ollama-b")

	sinkA := &fakeSink{}
	if err := adapterA.Execute(context.Background(), noderuntime.ExecutionSpec{
		RunID: "run-a", Target: "model-a", Input: map[string]any{"prompt": "test"},
	}, sinkA); err != nil {
		t.Fatalf("A execute: %v", err)
	}

	sinkB := &fakeSink{}
	if err := adapterB.Execute(context.Background(), noderuntime.ExecutionSpec{
		RunID: "run-b", Target: "model-b", Input: map[string]any{"prompt": "test"},
	}, sinkB); err != nil {
		t.Fatalf("B execute: %v", err)
	}

	if gotModelA != "model-a" {
		t.Errorf("A model: got %q, want model-a", gotModelA)
	}
	if gotNumCtxA != 4096 {
		t.Errorf("A num_ctx: got %d, want 4096", gotNumCtxA)
	}
	if gotModelB != "model-b" {
		t.Errorf("B model: got %q, want model-b", gotModelB)
	}
	if gotNumCtxB != 8192 {
		t.Errorf("B num_ctx: got %d, want 8192", gotNumCtxB)
	}

	capsA, _ := adapterA.Capabilities(context.Background())
	if capsA.InstanceKey != "ollama-a" {
		t.Errorf("A InstanceKey: got %q, want ollama-a", capsA.InstanceKey)
	}
	capsB, _ := adapterB.Capabilities(context.Background())
	if capsB.InstanceKey != "ollama-b" {
		t.Errorf("B InstanceKey: got %q, want ollama-b", capsB.InstanceKey)
	}
}

func TestOllamaContextSizeForcedOverRequestNumCtx(t *testing.T) {
	var gotNumCtx int
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		var req ollamaChatRequest
		if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
			t.Fatalf("decode request: %v", err)
		}
		if v, ok := req.Options["num_ctx"].(float64); ok {
			gotNumCtx = int(v)
		}
		w.Header().Set("Content-Type", "application/x-ndjson")
		_, _ = w.Write([]byte(`{"message":{"role":"assistant","content":"ok"},"done":false}` + "\n"))
		_, _ = w.Write([]byte(`{"done":true}` + "\n"))
	}))
	defer server.Close()

	// profile context_size=262144 must override request options.num_ctx=8192
	adapter := New(config.OllamaConf{BaseURL: server.URL, ContextSize: 262144}, zap.NewNop())
	sink := &fakeSink{}
	err := adapter.Execute(context.Background(), noderuntime.ExecutionSpec{
		RunID:  "run-forced",
		Target: "llama-test",
		Input: map[string]any{
			"prompt":  "hi",
			"options": map[string]any{"num_ctx": float64(8192), "temperature": 0.5},
		},
	}, sink)
	if err != nil {
		t.Fatalf("Execute failed: %v", err)
	}
	if gotNumCtx != 262144 {
		t.Fatalf("Edge-owned context_size must override request num_ctx: got %d, want 262144", gotNumCtx)
	}
}

func TestOllamaExecuteStreamsThinkingDeltasSeparately(t *testing.T) {
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if r.URL.Path != "/api/chat" {
			t.Fatalf("unexpected path %s", r.URL.Path)
		}
		w.Header().Set("Content-Type", "application/x-ndjson")
		_, _ = w.Write([]byte(`{"message":{"role":"assistant","thinking":"thinking..."},"done":false}` + "\n"))
		_, _ = w.Write([]byte(`{"message":{"role":"assistant","content":"answer"},"done":false}` + "\n"))
		_, _ = w.Write([]byte(`{"done":true,"prompt_eval_count":3,"eval_count":2}` + "\n"))
	}))
	defer server.Close()

	adapter := New(config.OllamaConf{BaseURL: server.URL}, zap.NewNop())
	sink := &fakeSink{}
	err := adapter.Execute(context.Background(), noderuntime.ExecutionSpec{
		RunID:  "run-thinking",
		Target: "llama-test",
		Input:  map[string]any{"prompt": "say hello"},
	}, sink)
	if err != nil {
		t.Fatalf("Execute failed: %v", err)
	}

	events := sink.all()
	if len(events) != 4 {
		t.Fatalf("expected 4 events, got %+v", events)
	}
	if events[0].Type != noderuntime.EventTypeStart {
		t.Fatalf("expected start event, got %+v", events[0])
	}
	if events[1].Type != noderuntime.EventTypeReasoningDelta || events[1].Delta != "thinking..." {
		t.Fatalf("expected reasoning_delta event, got %+v", events[1])
	}
	if events[2].Type != noderuntime.EventTypeDelta || events[2].Delta != "answer" {
		t.Fatalf("expected delta event, got %+v", events[2])
	}
	if events[3].Type != noderuntime.EventTypeComplete {
		t.Fatalf("expected complete event, got %+v", events[3])
	}
	if events[3].Usage == nil || events[3].Usage.InputTokens != 3 || events[3].Usage.OutputTokens != 2 {
		t.Fatalf("unexpected usage: %+v", events[3].Usage)
	}
}

func TestOllamaProbeProviderAvailability(t *testing.T) {
	t.Run("200_ok_and_target_hit", func(t *testing.T) {
		server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
			if r.URL.Path != "/api/tags" {
				t.Fatalf("unexpected path %s", r.URL.Path)
			}
			_, _ = w.Write([]byte(`{"models":[{"name":"llama-a"},{"name":"llama-b"}]}`))
		}))
		defer server.Close()

		adapter := New(config.OllamaConf{BaseURL: server.URL}, zap.NewNop())
		res, err := adapter.ProbeProvider(context.Background(), "llama-a")
		if err != nil {
			t.Fatalf("ProbeProvider failed: %v", err)
		}
		if res.Status != noderuntime.ProviderStatusAvailable {
			t.Errorf("expected Status available, got %s", res.Status)
		}
		if len(res.Targets) != 2 || res.Targets[0] != "llama-a" {
			t.Errorf("unexpected Targets: %+v", res.Targets)
		}
	})

	t.Run("200_ok_and_target_miss", func(t *testing.T) {
		server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
			_, _ = w.Write([]byte(`{"models":[{"name":"llama-a"}]}`))
		}))
		defer server.Close()

		adapter := New(config.OllamaConf{BaseURL: server.URL}, zap.NewNop())
		res, err := adapter.ProbeProvider(context.Background(), "llama-b")
		if err != nil {
			t.Fatalf("ProbeProvider failed: %v", err)
		}
		if res.Status != noderuntime.ProviderStatusUnavailable {
			t.Errorf("expected Status unavailable, got %s", res.Status)
		}
		if !strings.Contains(res.Detail, "not found") {
			t.Errorf("expected 'not found' in detail, got %s", res.Detail)
		}
	})

	t.Run("500_internal_error", func(t *testing.T) {
		server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
			w.WriteHeader(http.StatusInternalServerError)
		}))
		defer server.Close()

		adapter := New(config.OllamaConf{BaseURL: server.URL}, zap.NewNop())
		res, err := adapter.ProbeProvider(context.Background(), "llama-a")
		if err != nil {
			t.Fatalf("ProbeProvider failed: %v", err)
		}
		if res.Status != noderuntime.ProviderStatusUnavailable {
			t.Errorf("expected Status unavailable, got %s", res.Status)
		}
	})

	t.Run("empty_target", func(t *testing.T) {
		server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
			_, _ = w.Write([]byte(`{"models":[{"name":"llama-a"}]}`))
		}))
		defer server.Close()

		adapter := New(config.OllamaConf{BaseURL: server.URL}, zap.NewNop())
		res, err := adapter.ProbeProvider(context.Background(), "")
		if err != nil {
			t.Fatalf("ProbeProvider failed: %v", err)
		}
		if res.Status != noderuntime.ProviderStatusAvailable {
			t.Errorf("expected Status available, got %s", res.Status)
		}
	})
}