Vellum/internal/providers/tmdb_test.go

package providers

import (
	"context"
	"net/http"
	"net/http/httptest"
	"strings"
	"sync/atomic"
	"testing"
	"time"
)

// tmdbSrv answers both search endpoints from one handler, so a test can watch
// what each side did without two servers to keep in sync.
func tmdbSrv(t *testing.T, fn func(endpoint string, r *http.Request) (string, int)) *httptest.Server {
	t.Helper()
	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		body, status := fn(r.URL.Path, r)
		w.Header().Set("Content-Type", "application/json")
		w.WriteHeader(status)
		_, _ = w.Write([]byte(body))
	}))
	t.Cleanup(srv.Close)
	return srv
}

func TestTMDBMapsMovieAndTV(t *testing.T) {
	// Each endpoint records only its own request: the two handlers run on
	// different goroutines, so a shared capture would be a race.
	var gotMovieKey, gotMovieQuery, gotMovieYear, gotTVKey, gotTVYear string
	srv := tmdbSrv(t, func(endpoint string, r *http.Request) (string, int) {
		q := r.URL.Query()
		switch endpoint {
		case "/search/movie":
			gotMovieKey, gotMovieQuery, gotMovieYear = q.Get("api_key"), q.Get("query"), q.Get("year")
			return `{"results":[
				{"id": 603, "title": "The Matrix", "overview": "A hacker learns the truth.",
				 "release_date": "1999-03-31", "poster_path": "/matrix.jpg", "popularity": 88.1}
			]}`, http.StatusOK
		case "/search/tv":
			gotTVKey, gotTVYear = q.Get("api_key"), q.Get("first_air_date_year")
			return `{"results":[
				{"id": 1399, "name": "Game of Thrones", "overview": "Noble houses fight for a throne.",
				 "first_air_date": "2011-04-17", "poster_path": "/got.jpg", "popularity": 120.5}
			]}`, http.StatusOK
		}
		return `{"results":[]}`, http.StatusOK
	})

	p := NewTMDB("secret-key")
	p.BaseURL = srv.URL

	cands, err := p.Search(context.Background(), Query{Title: "The Matrix", Year: 1999})
	if err != nil {
		t.Fatalf("Search: %v", err)
	}
	if len(cands) != 2 {
		t.Fatalf("got %d candidates, want 2", len(cands))
	}

	// Merged by TMDB popularity: the show (120.5) outranks the film (88.1).
	show, movie := cands[0], cands[1]
	if show.Title != "Game of Thrones" || movie.Title != "The Matrix" {
		t.Fatalf("merge order wrong: %q then %q", show.Title, movie.Title)
	}

	if movie.Provider != "tmdb" || movie.ExternalID != "movie/603" {
		t.Errorf("movie id = %q (provider %q), want movie/603", movie.ExternalID, movie.Provider)
	}
	if movie.Year != 1999 {
		t.Errorf("movie Year = %d, want 1999", movie.Year)
	}
	if movie.Description != "A hacker learns the truth." {
		t.Errorf("movie Description = %q", movie.Description)
	}
	if movie.CoverURL != "https://image.tmdb.org/t/p/w500/matrix.jpg" {
		t.Errorf("movie CoverURL = %q", movie.CoverURL)
	}
	if show.ExternalID != "tv/1399" || show.Year != 2011 {
		t.Errorf("show id/year = %q/%d, want tv/1399/2011", show.ExternalID, show.Year)
	}
	if show.CoverURL != "https://image.tmdb.org/t/p/w500/got.jpg" {
		t.Errorf("show CoverURL = %q", show.CoverURL)
	}

	if gotMovieKey != "secret-key" || gotTVKey != "secret-key" {
		t.Errorf("api_key = movie %q, tv %q, want secret-key", gotMovieKey, gotTVKey)
	}
	if gotMovieQuery != "The Matrix" {
		t.Errorf("movie query = %q", gotMovieQuery)
	}
	// The two namespaces spell the year filter differently; both must get it.
	if gotMovieYear != "1999" || gotTVYear != "1999" {
		t.Errorf("year filter = movie %q, tv %q, want 1999", gotMovieYear, gotTVYear)
	}
}

func TestTMDBSearchesConcurrently(t *testing.T) {
	var arrived int32
	var overlapped atomic.Bool
	both := make(chan struct{})

	srv := tmdbSrv(t, func(_ string, _ *http.Request) (string, int) {
		// Neither endpoint answers until both requests are in flight, so a
		// sequential implementation deadlocks here rather than passing.
		if atomic.AddInt32(&arrived, 1) == 2 {
			overlapped.Store(true)
			close(both)
		}
		select {
		case <-both:
		case <-time.After(2 * time.Second):
		}
		return `{"results":[{"id": 1, "title": "Anything", "popularity": 1}]}`, http.StatusOK
	})

	p := NewTMDB("secret-key")
	p.BaseURL = srv.URL

	cands, err := p.Search(context.Background(), Query{Title: "Anything"})
	if err != nil {
		t.Fatalf("Search: %v", err)
	}
	if len(cands) != 2 {
		t.Errorf("got %d candidates, want one per endpoint", len(cands))
	}
	if n := atomic.LoadInt32(&arrived); n != 2 {
		t.Errorf("endpoints hit %d times, want 2", n)
	}
	if !overlapped.Load() {
		t.Error("the two searches did not overlap: they ran one after the other")
	}
}

// TMDB keeps films and television in one numeric id space: film 603 and show
// 603 are different works. A bare id would write them identically into
// provider_ids, so the namespace has to be part of what gets stored.
func TestTMDBExternalIDIsNamespaced(t *testing.T) {
	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		w.Header().Set("Content-Type", "application/json")
		// The same id from both endpoints is the collision under test.
		if strings.Contains(r.URL.Path, "/movie") {
			_, _ = w.Write([]byte(`{"results":[{"id":603,"title":"Film 603","release_date":"1999-03-31","popularity":10}]}`))
			return
		}
		_, _ = w.Write([]byte(`{"results":[{"id":603,"name":"Show 603","first_air_date":"2011-04-17","popularity":20}]}`))
	}))
	defer srv.Close()

	p := NewTMDB("secret-key")
	p.BaseURL = srv.URL

	cands, err := p.Search(context.Background(), Query{Title: "603"})
	if err != nil {
		t.Fatalf("Search: %v", err)
	}
	if len(cands) != 2 {
		t.Fatalf("got %d candidates, want 2", len(cands))
	}
	if cands[0].ExternalID == cands[1].ExternalID {
		t.Fatalf("a film and a show sharing id 603 both produced ExternalID %q", cands[0].ExternalID)
	}
}

func TestTMDBOneEndpointFailureKeepsOtherResults(t *testing.T) { // The movie endpoint only fails once the TV response has been written, so
	// this is an ordering guarantee, not a race the test happens to win.
	tvDone := make(chan struct{})
	srv := tmdbSrv(t, func(endpoint string, _ *http.Request) (string, int) {
		switch endpoint {
		case "/search/tv":
			close(tvDone)
			return `{"results":[{"id": 1396, "name": "Breaking Bad", "popularity": 200}]}`, http.StatusOK
		default:
			select {
			case <-tvDone:
			case <-time.After(2 * time.Second):
			}
			return `{"status_message":"boom"}`, http.StatusInternalServerError
		}
	})

	p := NewTMDB("secret-key")
	p.BaseURL = srv.URL

	cands, err := p.Search(context.Background(), Query{Title: "Breaking Bad"})
	if err != nil {
		t.Fatalf("Search: %v (one failing endpoint must not discard the other)", err)
	}
	if len(cands) != 1 || cands[0].Title != "Breaking Bad" {
		t.Fatalf("got %+v, want just the TV match", cands)
	}
}

func TestTMDBEmptyResultsAreNotAnError(t *testing.T) {
	srv := tmdbSrv(t, func(_ string, _ *http.Request) (string, int) {
		return `{"page":1,"results":[],"total_pages":1,"total_results":0}`, http.StatusOK
	})

	p := NewTMDB("secret-key")
	p.BaseURL = srv.URL

	cands, err := p.Search(context.Background(), Query{Title: "Nothing At All Like This"})
	if err != nil {
		t.Fatalf("Search: %v", err)
	}
	if len(cands) != 0 {
		t.Errorf("got %d candidates, want 0", len(cands))
	}
}

func TestTMDBWithoutUsableTitleDoesNotCallOut(t *testing.T) {
	var hits int32
	srv := tmdbSrv(t, func(_ string, _ *http.Request) (string, int) {
		atomic.AddInt32(&hits, 1)
		return `{"results":[{"id": 1, "title": "Unwanted", "popularity": 1}]}`, http.StatusOK
	})

	p := NewTMDB("secret-key")
	p.BaseURL = srv.URL

	for _, q := range []Query{{}, {Title: "   "}} {
		cands, err := p.Search(context.Background(), q)
		if err != nil {
			t.Fatalf("Search(%+v): %v", q, err)
		}
		if len(cands) != 0 {
			t.Errorf("Search(%+v) returned %d candidates, want 0", q, len(cands))
		}
	}
	if n := atomic.LoadInt32(&hits); n != 0 {
		t.Errorf("TMDB was called %d times with no usable title, want 0", n)
	}
}

func TestTMDBWithoutKeyDoesNothing(t *testing.T) {
	p := NewTMDB("")
	cands, err := p.Search(context.Background(), Query{Title: "The Matrix"})
	if err != nil {
		t.Fatalf("Search: %v", err)
	}
	if len(cands) != 0 {
		t.Errorf("got %d candidates with no key, want 0", len(cands))
	}
}