goyomi/sources/base/galleryadults/galleryadults.go

// Package galleryadults implements the GalleryAdults adult gallery base.
// HTML scraping; popular: GET {base}/language/{lang}/popular/?page={n}.
package galleryadults

import (
	"context"
	"fmt"
	"net/http"
	"regexp"
	"strings"

	"github.com/PuerkitoBio/goquery"

	"goyomi/internal/httpclient/flare"
	"goyomi/internal/source"
	"goyomi/sources/base/util"
)

type Config struct {
	Name      string
	BaseURL   string
	Lang      string
	MangaLang string // language path segment, e.g. "english"; empty = all
}

type Source struct {
	cfg    Config
	client *flare.Client
	id     int64
}

func New(cfg Config) *Source {
	c := flare.NewClient(flare.WithRateLimit(1, 2))
	return &Source{cfg: cfg, client: c, id: source.GenerateSourceID(cfg.Name, cfg.Lang)}
}

func (s *Source) ID() int64            { return s.id }
func (s *Source) Name() string         { return s.cfg.Name }
func (s *Source) Lang() string         { return s.cfg.Lang }
func (s *Source) SupportsLatest() bool { return true }

func (s *Source) base() string { return strings.TrimRight(s.cfg.BaseURL, "/") }

func (s *Source) get(ctx context.Context, rawURL string) (*goquery.Document, error) {
	resp, err := s.client.Get(ctx, rawURL)
	if err != nil {
		return nil, err
	}
	defer resp.Body.Close()
	if resp.StatusCode != http.StatusOK {
		return nil, fmt.Errorf("galleryadults: HTTP %d", resp.StatusCode)
	}
	return goquery.NewDocumentFromReader(resp.Body)
}

var shortenTitleRe = regexp.MustCompile(`(\[[^\]]*\]|[({][^)}]*[)}])`)
var tagCountRe = regexp.MustCompile(`\s*\(\d+\)\s*$`)

func (s *Source) mangaTitle(el *goquery.Selection, selector string) string {
	raw := strings.TrimSpace(el.Find(selector).Text())
	return strings.TrimSpace(shortenTitleRe.ReplaceAllString(raw, ""))
}

func imgAttr(img *goquery.Selection, baseURL string) string {
	for _, attr := range []string{"data-lazy-src", "data-src", "data-cfsrc", "src"} {
		if v, ok := img.Attr(attr); ok && v != "" && !strings.HasPrefix(v, "data:") {
			return util.AbsURL(baseURL, v)
		}
	}
	return ""
}

// thumbnailToFull converts thumbnail URLs to full image URLs by removing
// the "t" before the file extension (e.g. "1t.jpg" → "1.jpg").
func thumbnailToFull(u string) string {
	ext := u[strings.LastIndex(u, "."):]
	return strings.Replace(u, "t"+ext, ext, 1)
}

func (s *Source) mangaFromElement(el *goquery.Selection) source.SManga {
	m := source.SManga{}
	el.Find(".inner_thumb a").First().Each(func(_ int, a *goquery.Selection) {
		m.URL, _ = a.Attr("href")
	})
	m.Title = s.mangaTitle(el, ".caption")
	el.Find(".inner_thumb img").First().Each(func(_ int, img *goquery.Selection) {
		m.ThumbnailURL = imgAttr(img, s.cfg.BaseURL)
	})
	return m
}

func (s *Source) parsePage(doc *goquery.Document) source.MangasPage {
	var mangas []source.SManga
	doc.Find("div.thumb").Each(func(_ int, el *goquery.Selection) {
		m := s.mangaFromElement(el)
		if m.URL != "" && m.Title != "" {
			mangas = append(mangas, m)
		}
	})
	hasNext := doc.Find(".next.page-numbers, a[aria-label=Next]").Length() > 0
	return source.MangasPage{Mangas: mangas, HasNextPage: hasNext}
}

func (s *Source) langPath() string {
	if s.cfg.MangaLang != "" {
		return "language/" + s.cfg.MangaLang + "/"
	}
	return ""
}

func (s *Source) GetPopularManga(page int) (source.MangasPage, error) {
	u := fmt.Sprintf("%s/%spopular/?page=%d", s.base(), s.langPath(), page)
	doc, err := s.get(context.Background(), u)
	if err != nil {
		return source.MangasPage{}, err
	}
	return s.parsePage(doc), nil
}

func (s *Source) GetLatestUpdates(page int) (source.MangasPage, error) {
	u := fmt.Sprintf("%s/%s?page=%d", s.base(), s.langPath(), page)
	doc, err := s.get(context.Background(), u)
	if err != nil {
		return source.MangasPage{}, err
	}
	return s.parsePage(doc), nil
}

func (s *Source) GetSearchManga(page int, query string, filters []source.Filter) (source.MangasPage, error) {
	u := fmt.Sprintf("%s/search/?q=%s&page=%d", s.base(), query, page)
	doc, err := s.get(context.Background(), u)
	if err != nil {
		return source.MangasPage{}, err
	}
	return s.parsePage(doc), nil
}

// getInfo extracts tag links from elements matching ".tags" that contain "{tag}:".
func getInfo(el *goquery.Selection, tag string) string {
	var items []string
	el.Find(".tags").Each(func(_ int, tags *goquery.Selection) {
		if strings.Contains(tags.Text(), tag+":") {
			tags.Find("a.tag_btn").Each(func(_ int, a *goquery.Selection) {
				t := tagCountRe.ReplaceAllString(strings.TrimSpace(a.Text()), "")
				if t != "" {
					items = append(items, t)
				}
			})
		}
	})
	return strings.Join(items, ", ")
}

func (s *Source) GetMangaDetails(manga source.SManga) (source.SManga, error) {
	doc, err := s.get(context.Background(), util.AbsURL(s.cfg.BaseURL, manga.URL))
	if err != nil {
		return manga, err
	}
	result := source.SManga{URL: manga.URL, Status: source.StatusCompleted}
	top := doc.Find(".gallery_top")
	result.Title = s.mangaTitle(top, "h1")
	if result.Title == "" {
		result.Title = manga.Title
	}
	top.Find(".cover img").First().Each(func(_ int, img *goquery.Selection) {
		result.ThumbnailURL = imgAttr(img, s.cfg.BaseURL)
	})
	result.Genre = getInfo(top, "Tags")
	result.Author = getInfo(top, "Artists")
	return result, nil
}

func (s *Source) GetChapterList(manga source.SManga) ([]source.SChapter, error) {
	// Galleries have a single chapter: the gallery itself.
	return []source.SChapter{{
		URL:  manga.URL,
		Name: "Chapter",
	}}, nil
}

func (s *Source) GetPageList(chapter source.SChapter) ([]source.Page, error) {
	doc, err := s.get(context.Background(), util.AbsURL(s.cfg.BaseURL, chapter.URL))
	if err != nil {
		return nil, err
	}
	var pages []source.Page
	// Try thumbnail → full conversion first.
	doc.Find(".gallery_thumb a img").Each(func(i int, img *goquery.Selection) {
		if u := imgAttr(img, s.cfg.BaseURL); u != "" {
			pages = append(pages, source.Page{Index: i, ImageURL: thumbnailToFull(u)})
		}
	})
	if len(pages) == 0 {
		// Fallback: linked images directly.
		doc.Find(".gallery_thumb a").Each(func(i int, a *goquery.Selection) {
			if u, ok := a.Attr("href"); ok && u != "" {
				pages = append(pages, source.Page{Index: i, ImageURL: util.AbsURL(s.cfg.BaseURL, u)})
			}
		})
	}
	return pages, nil
}

func (s *Source) GetImageURL(page source.Page) (string, error) { return page.ImageURL, nil }
func (s *Source) GetFilterList() []source.Filter               { return nil }