fix: remove EPUB ad gaps and preserve chapter titles

This commit is contained in:
2026-07-17 23:55:22 +08:00
parent 1c5e11e8ef
commit 5b46c58262
2 changed files with 103 additions and 5 deletions
+13 -5
View File
@@ -266,8 +266,7 @@ func (b *Bilinovel) GetVolume(novelId int, volumeId int, skipChapterContent bool
if err != nil {
return nil, fmt.Errorf("failed to get chapter: %v", err)
}
chapter.Id = chapterId
volume.Chapters[i] = chapter
volume.Chapters[i] = mergeDownloadedChapter(volume.Chapters[i], chapter)
} else {
return nil, fmt.Errorf("failed to get chapter id: %v", volume.Chapters[i].Url)
}
@@ -277,6 +276,12 @@ func (b *Bilinovel) GetVolume(novelId int, volumeId int, skipChapterContent bool
return volume, nil
}
func mergeDownloadedChapter(listedChapter, downloadedChapter *model.Chapter) *model.Chapter {
downloadedChapter.Title = listedChapter.Title
downloadedChapter.Url = listedChapter.Url
return downloadedChapter
}
func (b *Bilinovel) getAllVolumes(novelId int, skipChapterContent bool, skipVolumes []int) ([]*model.Volume, error) {
b.logger.Info("Getting all volumes of novel", slog.Int("novelId", novelId))
@@ -442,9 +447,7 @@ func (b *Bilinovel) getChapterByPage(pwPage playwright.Page, chapter *model.Chap
chapter.Title = doc.Find("#atitle").Text()
}
content := doc.Find("#acontent").First()
content.Find(".cgo").Remove()
content.Find("center").Remove()
content.Find(".google-auto-placed").Remove()
cleanChapterContent(content)
if strings.Contains(resortedHtml, `font-family: "read"`) {
html, err := content.Find("p").Last().Html()
@@ -518,6 +521,11 @@ func (b *Bilinovel) getChapterByPage(pwPage playwright.Page, chapter *model.Chap
return hasNext, nil
}
func cleanChapterContent(content *goquery.Selection) {
content.Find(".cgo, .csgo, .co, .google-auto-placed, ins.adsbygoogle").Remove()
content.Find("center").Remove()
}
func (b *Bilinovel) getImg(url string) ([]byte, error) {
b.logger.Info("Getting img", slog.String("url", url))
resp, err := b.restyClient.R().SetHeader("Referer", "https://www.bilinovel.com").Get(url)