memby/server/internal/api/search_discover.go

package api

import (
	"context"
	"encoding/json"
	"net/http"
	"strconv"
	"strings"
	"sync"
	"time"

	"github.com/ponzischeme89/memby/server/internal/cache"
	"github.com/ponzischeme89/memby/server/internal/store"
)

// External discovery is a separate route from search on purpose.
//
// /v1/search answers out of Emby and Redis and is what a keystroke asks. This asks Sonarr
// and Radarr, which are somebody else's services on somebody else's network — a lookup
// measured in seconds, rate-limited upstream, and worth nothing at all for the letters on
// the way to a word. Splitting them is the whole of the promise that discovery can never
// delay the library: the television makes the two requests independently and renders
// whichever answers, so an *arr that is down or slow costs the second section and never
// the first.
const (
	// Three characters, not the search floor of two. Two letters match a large part of any
	// catalogue and the answer is noise; the *arrs are also the one backend here that Memby
	// does not own, so the floor is set by what is worth asking them rather than by what
	// they would tolerate.
	minDiscoveryQueryRunes = 3
	// How long one query's answer is reused for the whole household. A film's existence in
	// TMDb does not change minute to minute, and this is what makes leaving Search and
	// coming back, deleting and retyping a title, or a second television asking the same
	// thing free. It is deliberately well past the client's own debounce: the debounce
	// protects against typing, this protects against everything else.
	discoveryCacheTTL = 10 * time.Minute
	// A lookup that found nothing is kept for less. Nothing is the answer for a typo, which
	// is the query most likely to be corrected and asked again a second later — but it is
	// also the answer while an *arr is confused, and that recovers.
	discoveryEmptyTTL = 2 * time.Minute
	// Bounds the whole route. Longer than a library search, because that is the trade this
	// route exists to isolate, and short enough that a television is never left with a
	// section claiming it is still finding things for a minute.
	discoveryTimeout = 8 * time.Second
)

// discoveryResponse is deliberately not just an item list: a television has to be able to
// tell "there is nothing else to be had" from "we could not ask", because the first is an
// answer and the second is worth a quiet line under the results.
type discoveryResponse struct {
	Query string            `json:"query"`
	Items []json.RawMessage `json:"items"`
	// Partial means at least one source failed. The items that did arrive are still good.
	Partial bool `json:"partial,omitempty"`
	// Skipped names why nothing was asked at all, for the log and for the client's own
	// diagnostics. An empty list with no reason means the *arrs simply knew nothing.
	Skipped string `json:"skipped,omitempty"`
}

// normaliseDiscoveryQuery is what makes the cache and the client's own deduplication agree
// on when two searches are the same search. "Disclosure", "disclosure" and "disclosure  "
// are one question; "disclosure day" is another.
func normaliseDiscoveryQuery(term string) string {
	return strings.Join(strings.Fields(strings.ToLower(strings.TrimSpace(term))), " ")
}

// discoveryQueryEligible is the one rule, counted in runes rather than bytes for the reason
// searchQueryRecordable is: a title in Japanese is rejected at a third of an English one's
// length otherwise.
func discoveryQueryEligible(term string) bool {
	return len([]rune(normaliseDiscoveryQuery(term))) >= minDiscoveryQueryRunes
}

func (s *Server) handleSearchDiscover(w http.ResponseWriter, r *http.Request, sess store.Session) {
	started := time.Now()
	term := normaliseDiscoveryQuery(r.URL.Query().Get("q"))
	limit := queryInt(r, "limit", 20, 60)
	log := s.loggerFor(r.Context())

	if !discoveryQueryEligible(term) {
		s.writeDiscovery(w, discoveryResponse{Query: term, Skipped: "query_too_short"})
		return
	}
	sonarrOn := s.sonarrEnabled(r.Context())
	radarrOn := s.radarrEnabled(r.Context())
	if !sonarrOn && !radarrOn {
		s.writeDiscovery(w, discoveryResponse{Query: term, Skipped: "no_sources"})
		return
	}

	ctx, cancel := context.WithTimeout(r.Context(), discoveryTimeout)
	defer cancel()
	canRequest := s.requestAllowed(r, sess)

	type leg struct {
		source string
		items  []json.RawMessage
		cached bool
		err    error
		took   time.Duration
	}
	results := make(chan leg, 2)
	var wg sync.WaitGroup
	// Sonarr and Radarr are independent, and one being unreachable must never cost the
	// other's answer — the same reason the two halves of a request lookup run apart.
	start := func(source string, load func(context.Context) ([]json.RawMessage, error)) {
		wg.Add(1)
		go func() {
			defer wg.Done()
			at := time.Now()
			items, cached, err := s.discoverCached(ctx, source, term, limit, load)
			results <- leg{source: source, items: items, cached: cached, err: err, took: time.Since(at)}
		}()
	}
	if sonarrOn {
		start("sonarr", func(ctx context.Context) ([]json.RawMessage, error) {
			return s.discoverSonarr(ctx, term, limit)
		})
	}
	if radarrOn {
		start("radarr", func(ctx context.Context) ([]json.RawMessage, error) {
			return s.discoverRadarr(ctx, term, limit)
		})
	}
	go func() { wg.Wait(); close(results) }()

	merged := make([]json.RawMessage, 0, limit)
	response := discoveryResponse{Query: term}
	fields := []any{"query", term, "limit", limit}
	found := 0
	for result := range results {
		if result.err != nil {
			response.Partial = true
			log.Debug("discovery source failed", "query", term, "source", result.source, "error", result.err)
		}
		found += len(result.items)
		for _, item := range result.items {
			merged = mergeSearchRaw(merged, item, limit)
		}
		fields = append(fields,
			result.source, len(result.items),
			result.source+"_ms", result.took.Milliseconds(),
			result.source+"_cache", cacheWord(result.cached))
	}
	response.Items = applyRequestable(merged, canRequest)

	// DEBUG for the reason the search line is: this is the record of what somebody was
	// looking for, and it is a per-viewer stream of titles rather than something the
	// ordinary log should carry. An operator tuning the debounce turns it on.
	log.Debug("search discovery", append(fields,
		"results", len(response.Items),
		"duplicates", found-len(response.Items),
		"partial", response.Partial,
		"ms", time.Since(started).Milliseconds())...)
	s.writeDiscovery(w, response)
}

func (s *Server) writeDiscovery(w http.ResponseWriter, response discoveryResponse) {
	if response.Items == nil {
		response.Items = []json.RawMessage{}
	}
	writeJSON(w, http.StatusOK, response)
}

func cacheWord(hit bool) string {
	if hit {
		return "hit"
	}
	return "miss"
}

// discoverCached is where the cooldown lives.
//
// The key is a MetadataKey rather than a UserKey, and that is load-bearing: what Radarr
// knows about "disclosure day" is a fact about the world and the household's catalogue,
// with nothing in it derived from whoever typed it — so one lookup answers for every
// television in the house. Whether *this* viewer may press Request is applied afterwards,
// on the way out, which is the same separation cache.MetadataKey demands everywhere else.
//
// cachedRead brings the other half for free: two televisions missing the same key in the
// same second make one lookup between them rather than two.
func (s *Server) discoverCached(
	ctx context.Context,
	source, term string,
	limit int,
	load func(context.Context) ([]json.RawMessage, error),
) ([]json.RawMessage, bool, error) {
	key := cache.MetadataKey("discover:v1:" + source + ":" + itoa(limit) + ":" + term)
	body, hit, err := s.cachedRead(ctx, key, discoveryCacheTTL, func(ctx context.Context) (json.RawMessage, error) {
		items, err := load(ctx)
		if err != nil {
			return nil, err
		}
		return json.Marshal(nonNil(items))
	})
	if err != nil {
		return nil, hit, err
	}
	var items []json.RawMessage
	if err := json.Unmarshal(body, &items); err != nil {
		return nil, hit, err
	}
	return items, hit, nil
}

// discoverSonarr and discoverRadarr differ from the streaming search's own legs in one
// way: a title the household already has is dropped rather than listed. The library
// section above it is already showing that title, playable, and one result appearing twice
// under two headings is the single worst thing progressive search can do.
func (s *Server) discoverSonarr(ctx context.Context, term string, limit int) ([]json.RawMessage, error) {
	series, err := s.sonarr.Lookup(ctx, term)
	if err != nil {
		return nil, err
	}
	ids := make([]int, 0, len(series))
	for _, v := range series {
		if v.TVDBID > 0 {
			ids = append(ids, v.TVDBID)
		}
	}
	inLibrary, _ := s.store.LibraryContainsProviderIDs(ctx, "Tvdb", ids)
	items := make([]json.RawMessage, 0, limit)
	for _, v := range series {
		if v.TVDBID == 0 || inLibrary[v.TVDBID] || len(items) >= limit {
			continue
		}
		state := lookupStatusFor(RequestSubject{
			Tracked:  v.ID > 0,
			Released: seriesReleased(v.Status, v.NextAiring, time.Now()),
		}, false)
		items = append(items, searchExternalItem(v.Title, v.Year, v.Overview, "Series", "sonarr",
			state, strconv.Itoa(v.TVDBID), sonarrCoverURL(v.Images, "poster"), false))
	}
	return items, nil
}

func (s *Server) discoverRadarr(ctx context.Context, term string, limit int) ([]json.RawMessage, error) {
	movies, err := s.radarr.Lookup(ctx, term)
	if err != nil {
		return nil, err
	}
	ids := make([]int, 0, len(movies))
	for _, v := range movies {
		if v.TMDBID > 0 {
			ids = append(ids, v.TMDBID)
		}
	}
	inLibrary, _ := s.store.LibraryContainsProviderIDs(ctx, "Tmdb", ids)
	items := make([]json.RawMessage, 0, limit)
	for _, v := range movies {
		if v.TMDBID == 0 || inLibrary[v.TMDBID] || len(items) >= limit {
			continue
		}
		state := lookupStatusFor(RequestSubject{
			Tracked:  v.ID > 0,
			HasFile:  v.HasFile,
			Released: movieReleased(v.Status),
		}, false)
		items = append(items, searchExternalItem(v.Title, v.Year, v.Overview, "Movie", "radarr",
			state, strconv.Itoa(v.TMDBID), radarrCoverURL(v.Images, "poster"), false))
	}
	return items, nil
}

// applyRequestable is what keeps the cached half free of the viewer. Whether the button is
// offered is this person's permission and nothing else's, so it is stamped on the way out
// rather than baked into what the household shares.
func applyRequestable(items []json.RawMessage, canRequest bool) []json.RawMessage {
	for i, raw := range items {
		var item map[string]any
		if json.Unmarshal(raw, &item) != nil {
			continue
		}
		state, _ := item["MembySearchState"].(string)
		item["MembyRequestable"] = canRequest && state == RequestStatusRequestable
		if encoded, err := json.Marshal(item); err == nil {
			items[i] = encoded
		}
	}
	return items
}