goyomi/sources/base/madtheme/madtheme.go

// Package madtheme implements the MadTheme WordPress base.
// All list types via GET {base}/search?page={n}&sort=...
// Pages extracted from JSON blob in <script> tag.
package madtheme

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

	"github.com/PuerkitoBio/goquery"

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

type Config struct {
	Name    string
	BaseURL string
	Lang    string
}

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

func New(cfg Config) *Source {
	c := httpclient.NewClient(httpclient.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) get(ctx context.Context, rawURL string) (*goquery.Document, error) {
	req, err := http.NewRequestWithContext(ctx, http.MethodGet, rawURL, nil)
	if err != nil {
		return nil, err
	}
	req.Header.Set("Referer", s.cfg.BaseURL+"/")
	resp, err := s.client.Do(req)
	if err != nil {
		return nil, err
	}
	defer resp.Body.Close()
	if resp.StatusCode != http.StatusOK {
		return nil, fmt.Errorf("madtheme: HTTP %d", resp.StatusCode)
	}
	return goquery.NewDocumentFromReader(resp.Body)
}

func (s *Source) searchURL(page int, q, sort string) string {
	u, _ := url.Parse(strings.TrimRight(s.cfg.BaseURL, "/") + "/search")
	qv := u.Query()
	qv.Set("q", q)
	qv.Set("page", fmt.Sprintf("%d", page))
	if sort != "" {
		qv.Set("sort", sort)
	}
	u.RawQuery = qv.Encode()
	return u.String()
}

func (s *Source) parseMangaList(doc *goquery.Document) source.MangasPage {
	var mangas []source.SManga
	doc.Find("div.book-item, div.item, div.manga-item").Each(func(_ int, el *goquery.Selection) {
		m := source.SManga{}
		el.Find("a").First().Each(func(_ int, a *goquery.Selection) {
			if href, ok := a.Attr("href"); ok {
				m.URL = stripDomain(href, s.cfg.BaseURL)
			}
			m.Title = strings.TrimSpace(a.AttrOr("title", ""))
		})
		if m.Title == "" {
			el.Find("div.title, h3, h2").First().Each(func(_ int, e *goquery.Selection) {
				m.Title = strings.TrimSpace(e.Text())
			})
		}
		el.Find("img").First().Each(func(_ int, img *goquery.Selection) {
			m.ThumbnailURL = imgAttr(img, s.cfg.BaseURL)
		})
		if m.URL != "" {
			mangas = append(mangas, m)
		}
	})
	hasNext := doc.Find(".next, .pagination .next, a[rel=next]").Length() > 0
	return source.MangasPage{Mangas: mangas, HasNextPage: hasNext}
}

func (s *Source) GetPopularManga(page int) (source.MangasPage, error) {
	doc, err := s.get(context.Background(), s.searchURL(page, "", "views"))
	if err != nil {
		return source.MangasPage{}, err
	}
	return s.parseMangaList(doc), nil
}

func (s *Source) GetLatestUpdates(page int) (source.MangasPage, error) {
	doc, err := s.get(context.Background(), s.searchURL(page, "", "latest"))
	if err != nil {
		return source.MangasPage{}, err
	}
	return s.parseMangaList(doc), nil
}

func (s *Source) GetSearchManga(page int, query string, filters []source.Filter) (source.MangasPage, error) {
	doc, err := s.get(context.Background(), s.searchURL(page, query, ""))
	if err != nil {
		return source.MangasPage{}, err
	}
	return s.parseMangaList(doc), nil
}

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}
	doc.Find("h1, h2").First().Each(func(_ int, el *goquery.Selection) { result.Title = strings.TrimSpace(el.Text()) })
	doc.Find(".author a, .info a.author").First().Each(func(_ int, el *goquery.Selection) { result.Author = strings.TrimSpace(el.Text()) })
	doc.Find(".summary, .description, .manga-summary").First().Each(func(_ int, el *goquery.Selection) { result.Description = strings.TrimSpace(el.Text()) })
	doc.Find(".cover img, .manga-cover img").First().Each(func(_ int, img *goquery.Selection) { result.ThumbnailURL = imgAttr(img, s.cfg.BaseURL) })
	doc.Find(".status").First().Each(func(_ int, el *goquery.Selection) { result.Status = util.StatusFromString(el.Text()) })
	var genres []string
	doc.Find(".genres a, .genre a").Each(func(_ int, el *goquery.Selection) {
		if t := strings.TrimSpace(el.Text()); t != "" {
			genres = append(genres, t)
		}
	})
	result.Genre = strings.Join(genres, ", ")
	return result, nil
}

func (s *Source) GetChapterList(manga source.SManga) ([]source.SChapter, error) {
	doc, err := s.get(context.Background(), util.AbsURL(s.cfg.BaseURL, manga.URL))
	if err != nil {
		return nil, err
	}
	var chapters []source.SChapter
	doc.Find("ul.chapter-list li, .chapters li, .chapter-item").Each(func(i int, el *goquery.Selection) {
		ch := source.SChapter{}
		el.Find("a").First().Each(func(_ int, a *goquery.Selection) {
			if href, ok := a.Attr("href"); ok {
				ch.URL = stripDomain(href, s.cfg.BaseURL)
			}
			ch.Name = strings.TrimSpace(a.Text())
		})
		el.Find(".date, time").First().Each(func(_ int, e *goquery.Selection) {
			ch.DateUpload = util.ParseRelativeDate(e.Text())
		})
		if ch.URL != "" {
			chapters = append(chapters, ch)
		}
	})
	return chapters, nil
}

var pageJSONRe = regexp.MustCompile(`chapImages\s*=\s*'([^']+)'|"images"\s*:\s*(\[.*?])`)

func (s *Source) GetPageList(chapter source.SChapter) ([]source.Page, error) {
	rawURL := util.AbsURL(s.cfg.BaseURL, chapter.URL)
	doc, err := s.get(context.Background(), rawURL)
	if err != nil {
		return nil, err
	}
	var images []string
	doc.Find("script").Each(func(_ int, script *goquery.Selection) {
		text := script.Text()
		m := pageJSONRe.FindStringSubmatch(text)
		if len(m) > 1 {
			blob := m[1]
			if blob == "" {
				blob = m[2]
			}
			if blob != "" && images == nil {
				_ = json.Unmarshal([]byte(blob), &images)
			}
		}
	})
	if len(images) == 0 {
		doc.Find(".reading-content img, .chapter-content img").Each(func(i int, img *goquery.Selection) {
			if u := imgAttr(img, s.cfg.BaseURL); u != "" {
				images = append(images, u)
			}
		})
	}
	pages := make([]source.Page, len(images))
	for i, img := range images {
		pages[i] = source.Page{Index: i, URL: rawURL, ImageURL: util.AbsURL(s.cfg.BaseURL, img)}
	}
	return pages, nil
}

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

func stripDomain(href, baseURL string) string {
	parsed, err := url.Parse(href)
	if err != nil || !parsed.IsAbs() {
		return href
	}
	base, _ := url.Parse(baseURL)
	if base != nil && parsed.Host == base.Host {
		return parsed.RequestURI()
	}
	return href
}

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 ""
}