func _crawlAmebloPost(req *wcg.Request, post *hplink.AmebloPost, members []hplink.Member) error {
crawler := crawlers.NewAmebloPostCrawler(cacheutil.NewURLCacheAwareClient(req, entities.URLCache))
req.Logger.Infof("[Task.Crawlers.AmebloPosts] crawling URL: %s", post.URL)
crawled, err := crawler.RunOnPostURL(post.URL)
if err != nil {
return err
}
post.NumLikes = crawled.NumLikes
post.NumComments = crawled.NumComments
post.NumReblogs = crawled.NumReblogs
post.PostAt = crawled.PostAt
post.Title = crawled.Title
post.Theme = crawled.Theme
post.IsContentsCrawled = true
post.Images = make([]models.Image, len(crawled.ImageURLs))
if cacher, err := cacheutil.NewImageCacher(req, entities.ImageCache); err == nil {
iterator.ParallelSlice(crawled.ImageURLs, func(i int, url string) error {
req.Logger.Infof("[Task.Crawlers.AmebloPosts] caching image URL %s (on %s)", url, post.URL)
const retries = 5
var err error
var cache *models.ImageCache
for j := 0; j < retries; j++ {
if j != 0 {
req.Logger.Infof("[Task.Crawlers.AmebloPosts] Retry image URL caching: %s (on %s) (%d)", url, post.URL, j)
}
cache, err = cacher.Cache(url)
if err == nil {
post.Images[i] = *cache.ToImage()
return nil
}
lib.WaitAndEnsureAfter(lib.Now(), 3*time.Second)
}
req.Logger.Warnf("[Task.Crawlers.AmebloPosts] Failed to cache image URL %s (on %s) - %v", url, post.URL, err)
return err
})
} else {
for j, url := range crawled.ImageURLs {
req.Logger.Infof("[Task.Crawlers.AmebloPosts] skip caching image URL %s (on %s)", url, post.URL)
post.Images[j] = models.Image{
URL: url,
}
}
}
// Update MemberKey only if not set.
if post.MemberKey == "" {
if m := _guessMember(req, post, members); m != nil {
post.MemberKey = m.Key
}
}
req.Logger.Infof("[Task.Crawlers.AmebloPosts] finished crawling")
return nil
}
func _crawlAmebloEntryList(req *wcg.Request, settings *hplink.CrawlerSettings, url string) (string, error) {
_, _artist, err := entities.Artist.Get().Key(settings.ArtistKey).One(req)
if err != nil {
return "", err
}
p, err := entities.Member.Query().Filter("ArtistKey=", settings.ArtistKey).Execute(req)
if err != nil {
return "", err
}
if _artist == nil {
req.Logger.Warnf("No artist found by the key %q (CralwerSettings: %s)", settings.ArtistKey, settings.URL)
return "", nil
}
artist := _artist.(*hplink.Artist)
members := p.Data.([]hplink.Member)
crawler := crawlers.NewAmebloPostCrawler(urlfetch.NewHTTPClient(req))
postList, err := crawler.RunOnListURL(url)
if err != nil {
return "", fmt.Errorf("Error crawling %q - %v", url, err)
}
if len(postList.List) == 0 {
return "", nil
}
var amebloPostList []*hplink.AmebloPost
amebloPostKeys := make([]*datastore.Key, len(postList.List))
for i, post := range postList.List {
amebloPostKeys[i] = entities.AmebloPost.NewKey(req, post.URL, nil)
}
_, _currentPostList, err := entities.AmebloPost.GetMulti().DatastoreKeys(amebloPostKeys...).List(req)
if err != nil {
return "", err
}
currentPostList := _currentPostList.([]*hplink.AmebloPost)
for i, post := range postList.List {
if currentPostList[i] != nil && currentPostList[i].IsContentsCrawled {
// we don't update entities if they are already crawled by _processAmebloPosts
continue
}
amebloPost := &hplink.AmebloPost{
URL: post.URL,
Title: post.Title,
PostAt: post.PostAt,
Theme: post.Theme,
ArtistKey: artist.Key,
SettingsURL: settings.URL,
IsContentsCrawled: false,
}
if settings.MemberKey != "" {
amebloPost.MemberKey = settings.MemberKey
} else {
if m := _guessMember(req, amebloPost, members); m != nil {
amebloPost.MemberKey = m.Key
}
}
amebloPostList = append(amebloPostList, amebloPost)
}
_, err = entities.AmebloPost.PutMulti().DatastoreKeys(amebloPostKeys...).Update(req, amebloPostList)
if err != nil {
return "", err
}
return postList.Next, nil
}
