func TumblrParser(taskset *nandu.TaskSet, task *common.Task, bytes []byte) {
resp := new(TumblrResponse)
err := json.Unmarshal(bytes, resp)
if err != nil {
util.Error("failed to parse response %s\n", err.Error())
return
}
d := TaskTumblrData{}
task.GetData(&d)
if d.Sleep != 0 {
time.Sleep(time.Duration(d.Sleep) * time.Millisecond)
}
if d.Bid == 0 {
blog := getTumblrBlog(d.Name, taskset.GetDB())
d.Bid = int64(blog.ID)
}
util.Info("fetching %s\n", task.Url)
begin := int64(resp.Data.Blog.Posts) - d.Offset
end := begin - int64(len(resp.Data.Posts)) + 1
ibegin, iend := d.Update(begin, end)
for i := ibegin; i < iend; i++ {
post := resp.Data.Posts[i]
post.TumblrBlogID = uint(d.Bid)
post.Offset = uint(begin - i)
for j := range post.TumblrPhotos {
post.TumblrPhotos[j].Fill()
url := post.TumblrPhotos[j].Orig.Url
if fn, err := getFileName(url); err == nil {
util.Info("yield %s %s (%d | %d)\n", url, fn, resp.Data.Blog.Posts, begin-i)
}
}
taskset.GetDB().Create(&post)
}
if d.HasMore() {
new_task := new(common.Task)
new_task.Project = task.Project
new_task.TaskSet = task.TaskSet
d.Offset = int64(resp.Data.Blog.Posts) - d.Current + 1
new_task.SetData(d)
new_task.Url = genUrlFromInterval(&d)
taskset.GetWorker().Push(new_task)
}
}
func DownloadParser(taskset *nandu.TaskSet, task *common.Task, bs []byte) {
if gDownloadInfo == nil {
var err error
gDownloadInfo, err = NewDownloadInfo(kDownloadInfoFile)
if err != nil {
util.Fatal("%s\n", err.Error())
}
}
h := md5.New()
h.Write(bs)
hash := fmt.Sprintf("%x", h.Sum(nil))
data := DownloadData{}
task.GetData(&data)
if dup, fid := duplicate(taskset.GetDB(), hash); dup {
setFileId(taskset.GetDB(), data.Pid, fid)
return
}
iv, bs := encrypt(bs, gDownloadInfo.EncryptKey)
fileData := FileData{}
fileData.FileName = fmt.Sprintf("%x%s", iv[:4], hash[:24])
fileData.Hash = hash
err := gDownloadInfo.Bucket.PutObject(fileData.FileName, bytes.NewReader(bs))
if err != nil {
util.Error("can't create file %s\n", err.Error())
} else {
util.Info("downloading %s\n", task.Url)
taskset.GetDB().Create(&fileData)
setFileId(taskset.GetDB(), data.Pid, fileData.ID)
}
}