func ReadList(htmlsrc string) []Metadata {
entries := []Metadata{}
doc, err := html.Parse(strings.NewReader(htmlsrc))
if err != nil {
panic(err)
}
listNode := common.GetElementByClassName(doc, "gallery-content")
if listNode == nil {
return entries
}
for c := listNode.FirstChild; c != nil; c = c.NextSibling {
// 개행 노드는 쓸모없다
if len(strings.Trim(c.Data, "\n")) == 0 {
continue
}
if c.Type != html.ElementNode {
continue
}
metadata := ReadEntryNode(c)
entries = append(entries, metadata)
}
return entries
}
func readArtistNode(c *html.Node) []string {
artistParentNode := common.GetElementByClassName(c, "artist-list")
artistNodes := common.GetElementsByTagName(artistParentNode, "a")
artists := []string{}
for _, c := range artistNodes {
artist := c.FirstChild.Data
artists = append(artists, artist)
}
return artists
}
func readCoverNode(c *html.Node) []string {
coverParentNode := common.GetElementByClassName(c, "dj-img-cont")
coverNodes := common.GetElementsByTagName(coverParentNode, "img")
covers := []string{}
for _, c := range coverNodes {
cover := "https:" + c.Attr[0].Val
covers = append(covers, cover)
}
return covers
}
func readTagNode(c *html.Node) []string {
tagParentNode := common.GetElementByClassName(c, "relatedtags")
tagNodes := common.GetElementsByTagName(tagParentNode, "a")
tags := []string{}
for _, c := range tagNodes {
tag := c.FirstChild.Data
tag = sanitizeTag(tag)
tags = append(tags, tag)
}
return tags
}
func ReadEntryNode(n *html.Node) Metadata {
// language + type
// 특별한 구분자가 없어서 a 태그 전부 뽑은후 URL로 찾기
galleryType := ""
language := ""
series := []string{}
descNode := common.GetElementByClassName(n, "dj-desc")
aTags := common.GetElementsByTagName(descNode, "a")
// /type/doujinshi-all-1.html
typeRe := regexp.MustCompile(`/type/(.+)-(.+)-(\d+).html`)
// /index-korean-1.html
languageRe := regexp.MustCompile(`/index-(.+)-(\d+).html`)
// /series/kantai%20collection-all-1.html
seriesRe := regexp.MustCompile(`/series/(.+)-(.+)-(\d+).html`)
for _, c := range aTags {
if c.Attr[0].Key != "href" {
continue
}
url := c.Attr[0].Val
typeMatch := typeRe.FindStringSubmatch(url)
if typeMatch != nil {
galleryType = common.UrlDecode(typeMatch[1])
}
languageMatch := languageRe.FindStringSubmatch(url)
if languageMatch != nil {
language = common.UrlDecode(languageMatch[1])
}
seriesMatch := seriesRe.FindStringSubmatch(url)
if seriesMatch != nil {
series = append(series, common.UrlDecode(seriesMatch[1]))
}
}
return Metadata{
ID: readIdNode(n),
Title: readTitleNode(n),
Covers: readCoverNode(n),
Artists: readArtistNode(n),
Groups: []string{},
Type: galleryType,
Language: language,
Series: series,
Characters: []string{},
Tags: readTagNode(n),
Date: readDateNode(n),
}
}
func readDateNode(c *html.Node) string {
dateNode := common.GetElementByClassName(c, "date")
date := dateNode.FirstChild.Data
return date
}