func NewDatabase(t string, c string) *gorm.DB {
if t == "" || c == "" {
util.Error("db type and connect string can't be empty")
return nil
}
if database, err := gorm.Open(t, c); err != nil {
util.Error("can't connect to database %s\n", err.Error())
return nil
} else {
return &database
}
}
func (worker *Worker) Run() {
worker.validate()
util.Info("'%s' started\n", worker.project)
for {
task := worker.PopLocal()
if task == nil {
task = worker.Pop()
}
if task != nil {
worker.retryCount = 0
if taskset, ok := worker.tasksets[task.TaskSet]; !ok {
util.Error("can't find taskset %s\n", task.TaskSet)
continue
} else {
data := taskset.Fetch(task)
if data != nil {
taskset.Parse(task, data)
}
}
} else {
worker.retryCount += 1
if worker.retryCount >= worker.retryMax {
break
}
util.Info("sleep 1 second, ( %d | %d )\n", worker.retryCount, worker.retryMax)
time.Sleep(time.Second)
}
}
util.Info("'%s' exit\n", worker.project)
}
func (worker *Worker) Pop() *common.Task {
r, err := util.HttpPostJSON(worker.info.Server.AddrPop(), &common.Worker{worker.info.Server.Token, worker.project})
if err != nil {
util.Error("failed to pop task, %s\n", err.Error())
return nil
}
resp := new(common.CommonResponse)
err = util.HttpResponseUnmarshalJSON(resp, r, http.StatusOK)
if err != nil {
util.Error("failed to pop task, %s\n", err.Error())
return nil
} else {
if resp.Task != nil {
util.Debug(resp.Task.PopLog())
}
}
return resp.Task
}
func (worker *Worker) Push(task *common.Task) *common.Task {
task.Token = worker.info.Server.Token
task.Project = worker.project
r, err := util.HttpPostJSON(worker.info.Server.AddrPush(), task)
if err != nil {
util.Error("failed to push task, %s\n", err.Error())
return nil
}
resp := new(common.CommonResponse)
err = util.HttpResponseUnmarshalJSON(resp, r, http.StatusOK)
if err != nil {
util.Error("failed to push task, %s\n", err.Error())
return nil
} else {
if resp.Task != nil {
util.Debug(resp.Task.PushLog())
}
}
return resp.Task
}
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 CommonFetcher(taskset *TaskSet, task *common.Task) []byte {
r, err := taskset.client.Get(task.Url)
if err != nil {
util.Error("can't fetch url %s, %s\n", task.Url, err.Error())
return nil
}
defer r.Body.Close()
bytes, err := ioutil.ReadAll(r.Body)
if err != nil {
util.Error("can't get bytes from url %s, %s\n", task.Url, err.Error())
return nil
}
if r.StatusCode/100 != 2 {
util.Error("get wrong status code url %s, %d\n", task.Url, r.StatusCode)
return nil
}
return bytes
}
func NewOauth(appKey string, appSecret string, token string, secret string) *http.Client {
if token == "" || secret == "" || appKey == "" || appSecret == "" {
util.Error("token can't be empty\n")
return nil
}
consumer := oauth.NewConsumer(
appKey,
appSecret,
oauth.ServiceProvider{
"", "", "", "",
})
t := new(oauth.AccessToken)
t.Token = token
t.Secret = secret
client, err := consumer.MakeHttpClient(t)
if err != nil {
util.Error("%s\n", err.Error())
return nil
}
return client
}
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)
}
}