kafka/internal/engines/wikipedia.go

package engines

import (
	"context"
	"encoding/json"
	"errors"
	"fmt"
	"io"
	"net/http"
	"net/url"
	"strings"

	"github.com/ashie/gosearch/internal/contracts"
)

type WikipediaEngine struct {
	client *http.Client
}

func (e *WikipediaEngine) Name() string { return "wikipedia" }

func (e *WikipediaEngine) Search(ctx context.Context, req contracts.SearchRequest) (contracts.SearchResponse, error) {
	if e == nil || e.client == nil {
		return contracts.SearchResponse{}, errors.New("wikipedia engine not initialized")
	}
	if strings.TrimSpace(req.Query) == "" {
		return contracts.SearchResponse{Query: req.Query}, nil
	}

	lang := strings.TrimSpace(req.Language)
	if lang == "" || lang == "auto" {
		lang = "en"
	}
	// Wikipedia subdomains are based on the language code; keep it simple for MVP.
	lang = strings.SplitN(lang, "-", 2)[0]
	lang = strings.ReplaceAll(lang, "_", "-")
	wikiNetloc := fmt.Sprintf("%s.wikipedia.org", lang)

	endpoint := fmt.Sprintf(
		"https://%s/api/rest_v1/page/summary/%s",
		wikiNetloc,
		url.PathEscape(req.Query),
	)

	httpReq, err := http.NewRequestWithContext(ctx, http.MethodGet, endpoint, nil)
	if err != nil {
		return contracts.SearchResponse{}, err
	}
	// Wikimedia APIs require a descriptive User-Agent.
	httpReq.Header.Set(
		"User-Agent",
		"gosearch-go/0.1 (compatible; +https://github.com/ashie/gosearch)",
	)
	// Best-effort: hint content language.
	if req.Language != "" && req.Language != "auto" {
		httpReq.Header.Set("Accept-Language", req.Language)
	}

	resp, err := e.client.Do(httpReq)
	if err != nil {
		return contracts.SearchResponse{}, err
	}
	defer resp.Body.Close()

	if resp.StatusCode == http.StatusNotFound {
		return contracts.SearchResponse{
			Query:            req.Query,
			NumberOfResults: 0,
			Results:          []contracts.MainResult{},
			Answers:          []map[string]any{},
			Corrections:      []string{},
			Infoboxes:        []map[string]any{},
			Suggestions:      []string{},
			UnresponsiveEngines: [][2]string{},
		}, nil
	}
	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
		body, _ := io.ReadAll(io.LimitReader(resp.Body, 16*1024))
		return contracts.SearchResponse{}, fmt.Errorf("wikipedia upstream error: status=%d body=%q", resp.StatusCode, string(body))
	}

	var api struct {
		Title       string `json:"title"`
		Description string `json:"description"`
		Titles      struct {
			Display string `json:"display"`
		} `json:"titles"`
		ContentURLs struct {
			Desktop struct {
				Page string `json:"page"`
			} `json:"desktop"`
		} `json:"content_urls"`
	}

	if err := json.NewDecoder(resp.Body).Decode(&api); err != nil {
		return contracts.SearchResponse{}, err
	}

	pageURL := api.ContentURLs.Desktop.Page
	if pageURL == "" {
		// API returned a non-standard payload; treat as no result.
		return contracts.SearchResponse{
			Query:               req.Query,
			NumberOfResults:    0,
			Results:             []contracts.MainResult{},
			Answers:             []map[string]any{},
			Corrections:         []string{},
			Infoboxes:           []map[string]any{},
			Suggestions:         []string{},
			UnresponsiveEngines: [][2]string{},
		}, nil
	}

	title := api.Titles.Display
	if title == "" {
		title = api.Title
	}

	content := api.Description

	urlPtr := pageURL
	pub := (*string)(nil)

	results := []contracts.MainResult{
		{
			Template: "default.html",
			Title:    title,
			Content:  content,
			URL:      &urlPtr,
			Pubdate:  pub,
			Engine:   "wikipedia",
			Score:    0,
			Category: "general",
			Priority: "",
			Positions: nil,
			Engines:  []string{"wikipedia"},
		},
	}

	return contracts.SearchResponse{
		Query:               req.Query,
		NumberOfResults:    len(results),
		Results:             results,
		Answers:             []map[string]any{},
		Corrections:         []string{},
		Infoboxes:           []map[string]any{},
		Suggestions:         []string{},
		UnresponsiveEngines: [][2]string{},
	}, nil
}