canon/internal/providers/wikipedia.go

package providers

import (
	"context"
	"net/url"
	"regexp"
	"sort"
	"strconv"
	"strings"
)

// wikipedia finds stage works (musicals, plays, operas) on English
// Wikipedia. There's no free, open theater database with an API, and
// Wikipedia has the best coverage plus poster/logo art for most shows. The
// specific production (Broadway, a tour, …) is chosen by the owner.
type wikipedia struct {
	r *Registry
	// stage limits results to musicals, plays and operas. Custom lists
	// search all of Wikipedia.
	stage bool
}

func (w *wikipedia) Name() string  { return "wikipedia" }
func (w *wikipedia) Label() string { return "Wikipedia" }

const theaterTemplates = `hastemplate:"Infobox musical|Infobox play|Infobox opera"`

func (w *wikipedia) Search(ctx context.Context, query string) ([]Result, error) {
	if !w.stage {
		return w.search(ctx, query)
	}
	res, err := w.search(ctx, query+" "+theaterTemplates)
	if err == nil && len(res) == 0 {
		res, err = w.search(ctx, query)
	}
	return res, err
}

var leadingYear = regexp.MustCompile(`^(1[5-9]\d\d|20\d\d)\s+`)

func (w *wikipedia) search(ctx context.Context, query string) ([]Result, error) {
	return w.pages(ctx, url.Values{"generator": {"search"}, "gsrsearch": {query}, "gsrlimit": {"15"}})
}

// Lookup refreshes one page by its page ID.
func (w *wikipedia) Lookup(ctx context.Context, providerID string) (Result, error) {
	res, err := w.pages(ctx, url.Values{"pageids": {providerID}})
	if err != nil {
		return Result{}, err
	}
	if len(res) == 0 {
		return Result{}, ErrGone
	}
	return res[0], nil
}

// pages runs a query for pages (a search, or specific page IDs) and shapes
// them into results.
func (w *wikipedia) pages(ctx context.Context, which url.Values) ([]Result, error) {
	q := url.Values{
		"action":        {"query"},
		"format":        {"json"},
		"formatversion": {"2"},
		"prop":          {"pageimages|description|info|extracts"},
		"piprop":        {"thumbnail"},
		"pithumbsize":   {"800"},
		"pilicense":     {"any"},
		"inprop":        {"url"},
		"exintro":       {"1"},
		"explaintext":   {"1"},
		"exsentences":   {"3"},
		"exlimit":       {"max"},
	}
	for k, v := range which {
		q[k] = v
	}
	var resp struct {
		Query struct {
			Pages []struct {
				PageID      int64  `json:"pageid"`
				Missing     bool   `json:"missing"`
				Title       string `json:"title"`
				Index       int    `json:"index"`
				Description string `json:"description"`
				Extract     string `json:"extract"`
				FullURL     string `json:"fullurl"`
				Thumbnail   struct {
					Source string `json:"source"`
				} `json:"thumbnail"`
			} `json:"pages"`
		} `json:"query"`
	}
	if err := w.r.getJSON(ctx, "https://en.wikipedia.org/w/api.php?"+q.Encode(), nil, &resp); err != nil {
		return nil, err
	}
	pages := resp.Query.Pages
	sort.Slice(pages, func(i, j int) bool { return pages[i].Index < pages[j].Index })
	out := make([]Result, 0, len(pages))
	for _, p := range pages {
		if p.Missing {
			continue
		}
		title := p.Title
		// "Hamilton (musical)" → "Hamilton"
		if i := strings.LastIndex(title, " ("); i > 0 && strings.HasSuffix(title, ")") {
			title = title[:i]
		}
		desc := leadingYear.ReplaceAllString(p.Description, "")
		if desc != "" {
			desc = strings.ToUpper(desc[:1]) + desc[1:]
		}
		img := p.Thumbnail.Source
		if i := strings.Index(img, "?"); i > 0 {
			img = img[:i]
		}
		out = append(out, Result{
			Provider:    "wikipedia",
			ProviderID:  strconv.FormatInt(p.PageID, 10),
			Title:       title,
			Subtitle:    desc,
			Year:        yearOf(p.Description),
			ImageURL:    img,
			ExternalURL: p.FullURL,
			Overview:    truncate(p.Extract, maxDescription),
		})
	}
	return out, nil
}