Feature 10: Recipe grounding and photos (#78)

Co-authored-by: dev_agent <dev_agent@cubecraftlabs.com>

internal/recipes/search.go

@@ -0,0 +1,314 @@
+package recipes
+
+import (
+	"context"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io"
+	"net/http"
+	"net/url"
+	"strconv"
+	"strings"
+	"time"
+
+	"code.cubecraftlabs.com/Maniacs_Mansion/Meal-Quest/internal/db"
+	"code.cubecraftlabs.com/Maniacs_Mansion/Meal-Quest/internal/secretbox"
+)
+
+const (
+	// MaxSearchResults is the most candidates one search may ask for; it is
+	// Brave's largest page.
+	MaxSearchResults = 20
+	// SearchTimeout bounds one search request, reading the body included.
+	SearchTimeout = 10 * time.Second
+	// MaxSearchBodyBytes caps a search response body. A larger one is an
+	// error, never truncated.
+	MaxSearchBodyBytes = 1 << 20
+	// BraveEndpoint is the Brave Web Search API.
+	BraveEndpoint = "https://api.search.brave.com/res/v1/web/search"
+)
+
+var (
+	// ErrSearchRefused is returned for HTTP 401 or 403: Brave rejected the
+	// API key, or SearXNG does not allow format=json.
+	ErrSearchRefused = errors.New("recipes: search provider refused the request")
+	// ErrSearchRateLimited is returned for HTTP 429.
+	ErrSearchRateLimited = errors.New("recipes: search provider rate limit reached")
+	// ErrSearchResponse is returned for a response body that is not the
+	// provider's JSON shape — SearXNG without a results array, Brave without
+	// "type": "search" or with a web object lacking a results array — is
+	// over MaxSearchBodyBytes, or holds a result without an absolute http(s)
+	// URL.
+	ErrSearchResponse = errors.New("recipes: malformed search response")
+)
+
+// SearchStatusError is returned for any other non-2xx search response,
+// redirects included: a search endpoint is called directly and never
+// followed elsewhere.
+type SearchStatusError struct {
+	Provider string
+	Status   int
+}
+
+func (e *SearchStatusError) Error() string {
+	return fmt.Sprintf("recipes: %s search: HTTP %d", e.Provider, e.Status)
+}
+
+// Candidate is one search result: a page that may hold a recipe.
+type Candidate struct {
+	URL     string
+	Title   string
+	Snippet string
+}
+
+// SearchProvider finds recipe pages on the web.
+type SearchProvider interface {
+	// Search returns up to n candidates for query, best first. The query
+	// gets " recipe" appended unless it already has the word. n must be
+	// 1 to MaxSearchResults.
+	Search(ctx context.Context, query string, n int) ([]Candidate, error)
+}
+
+// NewSearchProvider returns the provider s names. SearXNG is called at
+// s.BaseUrl; Brave is called with the API key in s.ApiKeyCt, decrypted by
+// box.
+func NewSearchProvider(s db.RecipeSearchSetting, box *secretbox.Box) (SearchProvider, error) {
+	switch s.Provider {
+	case "searxng":
+		return NewSearXNG(s.BaseUrl.String)
+	case "brave":
+		if len(s.ApiKeyCt) == 0 {
+			return nil, errors.New("recipes: Brave search has no API key")
+		}
+		key, err := box.Decrypt(s.ApiKeyCt)
+		if err != nil {
+			return nil, fmt.Errorf("recipes: Brave API key: %w", err)
+		}
+		return NewBrave(string(key))
+	default:
+		return nil, fmt.Errorf("recipes: unknown search provider %q", s.Provider)
+	}
+}
+
+// SearXNG searches through a SearXNG instance. Its base URL is a configured
+// endpoint and may be on a private network, so it is called directly rather
+// than through the public-only Fetcher.
+type SearXNG struct {
+	endpoint *url.URL
+	client   *http.Client
+}
+
+// NewSearXNG returns a SearXNG provider for the instance at baseURL, an
+// absolute http:// or https:// URL without credentials, query or fragment.
+func NewSearXNG(baseURL string) (*SearXNG, error) {
+	if baseURL == "" {
+		return nil, errors.New("recipes: SearXNG URL is not set")
+	}
+	u, err := url.Parse(baseURL)
+	if err != nil || (u.Scheme != "http" && u.Scheme != "https") || u.Host == "" {
+		return nil, errors.New("recipes: SearXNG URL must be an absolute http:// or https:// URL")
+	}
+	if u.User != nil || u.RawQuery != "" || u.Fragment != "" {
+		return nil, errors.New("recipes: SearXNG URL must not have credentials, a query or a fragment")
+	}
+	return &SearXNG{endpoint: u.JoinPath("search"), client: newSearchClient()}, nil
+}
+
+// Search implements SearchProvider with
+// {base}/search?q=…&format=json&categories=general.
+func (s *SearXNG) Search(ctx context.Context, query string, n int) ([]Candidate, error) {
+	q, err := searchQuery(query, n)
+	if err != nil {
+		return nil, err
+	}
+	u := *s.endpoint
+	u.RawQuery = url.Values{"q": {q}, "format": {"json"}, "categories": {"general"}}.Encode()
+	req, err := http.NewRequestWithContext(ctx, http.MethodGet, u.String(), nil)
+	if err != nil {
+		return nil, fmt.Errorf("recipes: build SearXNG request: %w", err)
+	}
+	// Results is a pointer so a body without the array is told apart from
+	// an empty search.
+	var body struct {
+		Results *[]struct {
+			URL     string `json:"url"`
+			Title   string `json:"title"`
+			Content string `json:"content"`
+		} `json:"results"`
+	}
+	if err := doSearch(s.client, req, "SearXNG", &body); err != nil {
+		return nil, err
+	}
+	if body.Results == nil {
+		return nil, fmt.Errorf("%w: SearXNG body has no results array", ErrSearchResponse)
+	}
+	var out []Candidate
+	for _, r := range *body.Results {
+		out = append(out, Candidate{URL: r.URL, Title: r.Title, Snippet: r.Content})
+	}
+	return candidates("SearXNG", out, n)
+}
+
+// Brave searches through the Brave Web Search API.
+type Brave struct {
+	key      string
+	endpoint string
+	client   *http.Client
+}
+
+// NewBrave returns a Brave provider that authenticates with key, the
+// decrypted API key.
+func NewBrave(key string) (*Brave, error) {
+	if key == "" {
+		return nil, errors.New("recipes: Brave API key is empty")
+	}
+	for i := 0; i < len(key); i++ {
+		if key[i] < 0x21 || key[i] > 0x7e {
+			// The key is never echoed.
+			return nil, errors.New("recipes: Brave API key has a character not allowed in a header")
+		}
+	}
+	return &Brave{key: key, endpoint: BraveEndpoint, client: newSearchClient()}, nil
+}
+
+// Search implements SearchProvider with GET BraveEndpoint?q=…&count=n.
+func (b *Brave) Search(ctx context.Context, query string, n int) ([]Candidate, error) {
+	q, err := searchQuery(query, n)
+	if err != nil {
+		return nil, err
+	}
+	u := b.endpoint + "?" + url.Values{"q": {q}, "count": {strconv.Itoa(n)}}.Encode()
+	req, err := http.NewRequestWithContext(ctx, http.MethodGet, u, nil)
+	if err != nil {
+		return nil, fmt.Errorf("recipes: build Brave request: %w", err)
+	}
+	req.Header.Set("X-Subscription-Token", b.key)
+	// Brave leaves out "web" when nothing matched, so a search response is
+	// recognised by its "type"; a web object present must hold the results
+	// array.
+	var body struct {
+		Type string `json:"type"`
+		Web  *struct {
+			Results *[]struct {
+				URL         string `json:"url"`
+				Title       string `json:"title"`
+				Description string `json:"description"`
+			} `json:"results"`
+		} `json:"web"`
+	}
+	if err := doSearch(b.client, req, "Brave", &body); err != nil {
+		return nil, err
+	}
+	if body.Type != "search" {
+		return nil, fmt.Errorf("%w: Brave body type is %q, not \"search\"", ErrSearchResponse, body.Type)
+	}
+	if body.Web == nil {
+		return nil, nil
+	}
+	if body.Web.Results == nil {
+		return nil, fmt.Errorf("%w: Brave web object has no results array", ErrSearchResponse)
+	}
+	var out []Candidate
+	for _, r := range *body.Web.Results {
+		out = append(out, Candidate{URL: r.URL, Title: r.Title, Snippet: r.Description})
+	}
+	return candidates("Brave", out, n)
+}
+
+func newSearchClient() *http.Client {
+	return &http.Client{
+		Timeout: SearchTimeout,
+		CheckRedirect: func(*http.Request, []*http.Request) error {
+			return http.ErrUseLastResponse
+		},
+	}
+}
+
+// searchQuery validates n and returns query with " recipe" appended unless
+// one of its words already is "recipe", in any case.
+func searchQuery(query string, n int) (string, error) {
+	if n < 1 || n > MaxSearchResults {
+		return "", fmt.Errorf("recipes: search result count %d is outside 1-%d", n, MaxSearchResults)
+	}
+	q := strings.TrimSpace(query)
+	if q == "" {
+		return "", errors.New("recipes: search query is empty")
+	}
+	for _, w := range strings.Fields(q) {
+		if strings.EqualFold(w, "recipe") {
+			return q, nil
+		}
+	}
+	return q + " recipe", nil
+}
+
+// doSearch sends req and decodes a 2xx JSON body into v, mapping failures to
+// the search errors.
+func doSearch(client *http.Client, req *http.Request, provider string, v any) error {
+	req.Header.Set("Accept", "application/json")
+	req.Header.Set("User-Agent", UserAgent)
+	resp, err := client.Do(req)
+	if err != nil {
+		if ctxErr := req.Context().Err(); ctxErr != nil {
+			return ctxErr
+		}
+		return fmt.Errorf("recipes: %s search: %w", provider, err)
+	}
+	defer resp.Body.Close()
+	switch {
+	case resp.StatusCode == http.StatusUnauthorized || resp.StatusCode == http.StatusForbidden:
+		return fmt.Errorf("%w: %s HTTP %d", ErrSearchRefused, provider, resp.StatusCode)
+	case resp.StatusCode == http.StatusTooManyRequests:
+		return fmt.Errorf("%w: %s", ErrSearchRateLimited, provider)
+	case resp.StatusCode < 200 || resp.StatusCode > 299:
+		return &SearchStatusError{Provider: provider, Status: resp.StatusCode}
+	}
+	data, err := io.ReadAll(io.LimitReader(resp.Body, MaxSearchBodyBytes+1))
+	if err != nil {
+		return fmt.Errorf("recipes: %s search: read body: %w", provider, err)
+	}
+	if len(data) > MaxSearchBodyBytes {
+		return fmt.Errorf("%w: %s body is over %d bytes", ErrSearchResponse, provider, MaxSearchBodyBytes)
+	}
+	if err := json.Unmarshal(data, v); err != nil {
+		return fmt.Errorf("%w: %s: %v", ErrSearchResponse, provider, err)
+	}
+	return nil
+}
+
+// candidates cleans results — HTML tags and entities, which both providers
+// leave in titles and snippets, are removed — checks every URL is absolute
+// http(s) and keeps the first n.
+func candidates(provider string, results []Candidate, n int) ([]Candidate, error) {
+	if len(results) > n {
+		results = results[:n]
+	}
+	out := make([]Candidate, 0, len(results))
+	for i, r := range results {
+		u, err := url.Parse(r.URL)
+		if err != nil || (u.Scheme != "http" && u.Scheme != "https") || u.Host == "" {
+			return nil, fmt.Errorf("%w: %s result %d has no absolute http(s) URL", ErrSearchResponse, provider, i)
+		}
+		out = append(out, Candidate{URL: r.URL, Title: stripTags(r.Title), Snippet: stripTags(r.Snippet)})
+	}
+	return out, nil
+}
+
+// stripTags drops HTML tags such as Brave's <strong> highlights, then
+// decodes entities and trims space.
+func stripTags(s string) string {
+	var b strings.Builder
+	inTag := false
+	for _, r := range s {
+		switch {
+		case r == '<':
+			inTag = true
+		case r == '>' && inTag:
+			inTag = false
+		case !inTag:
+			b.WriteRune(r)
+		}
+	}
+	return cleanText(b.String())
+}