Merge pull request #357 from majestrate/scraper

fixes

common/search.go

@@ -19,6 +19,9 @@ const (
 	Date
 	Downloads
 	Size
+	Seeders
+	Leechers
+	Completed
 )
 
 type Category struct {
@@ -44,5 +47,6 @@ type SearchParam struct {
 	Page     int
 	UserID   uint
 	Max      uint
+	NotNull  string
 	Query    string
 }

main.go

@@ -84,6 +84,7 @@ func RunScraper(conf *config.Config) {
 	signals.RegisterCloser(scraper)
 	// run udp scraper worker
 	for workers > 0 {
+		log.Infof("starting up worker %d", workers)
 		go scraper.RunWorker(pc)
 		workers--
 	}

router/modpanel.go

@@ -3,9 +3,7 @@ package router
 import (
 	"fmt"
 	"html"
-	"html/template"
 	"net/http"
-	"path/filepath"
 	"strconv"
 
 	"github.com/ewhal/nyaa/db"
@@ -23,23 +21,6 @@ import (
 	"github.com/gorilla/mux"
 )
 
-var panelIndex, panelTorrentList, panelUserList, panelCommentList, panelTorrentEd, panelTorrentReportList *template.Template
-
-func init() {
-	panelTorrentList = template.Must(template.New("torrentlist").Funcs(FuncMap).ParseFiles(filepath.Join(TemplateDir, "admin_index.html"), filepath.Join(TemplateDir, "admin/torrentlist.html")))
-	panelTorrentList = template.Must(panelTorrentList.ParseGlob(filepath.Join(TemplateDir, "_*.html")))
-	panelUserList = template.Must(template.New("userlist").Funcs(FuncMap).ParseFiles(filepath.Join(TemplateDir, "admin_index.html"), filepath.Join(TemplateDir, "admin/userlist.html")))
-	panelUserList = template.Must(panelUserList.ParseGlob(filepath.Join(TemplateDir, "_*.html")))
-	panelCommentList = template.Must(template.New("commentlist").Funcs(FuncMap).ParseFiles(filepath.Join(TemplateDir, "admin_index.html"), filepath.Join(TemplateDir, "admin/commentlist.html")))
-	panelCommentList = template.Must(panelCommentList.ParseGlob(filepath.Join(TemplateDir, "_*.html")))
-	panelIndex = template.Must(template.New("indexPanel").Funcs(FuncMap).ParseFiles(filepath.Join(TemplateDir, "admin_index.html"), filepath.Join(TemplateDir, "admin/panelindex.html")))
-	panelIndex = template.Must(panelIndex.ParseGlob(filepath.Join(TemplateDir, "_*.html")))
-	panelTorrentEd = template.Must(template.New("torrent_ed").Funcs(FuncMap).ParseFiles(filepath.Join(TemplateDir, "admin_index.html"), filepath.Join(TemplateDir, "admin/paneltorrentedit.html")))
-	panelTorrentEd = template.Must(panelTorrentEd.ParseGlob(filepath.Join(TemplateDir, "_*.html")))
-	panelTorrentReportList = template.Must(template.New("torrent_report").Funcs(FuncMap).ParseFiles(filepath.Join(TemplateDir, "admin_index.html"), filepath.Join(TemplateDir, "admin/torrent_report.html")))
-	panelTorrentReportList = template.Must(panelTorrentReportList.ParseGlob(filepath.Join(TemplateDir, "_*.html")))
-}
-
 func IndexModPanel(w http.ResponseWriter, r *http.Request) {
 	currentUser := GetUser(r)
 	if userPermission.HasAdmin(currentUser) {
@@ -217,12 +198,12 @@ func TorrentPostEditModPanel(w http.ResponseWriter, r *http.Request) {
 	err := form.NewErrors()
 	infos := form.NewInfos()
 	torrent, _ := torrentService.GetTorrentById(id)
-	if (torrent.ID > 0) {
+	if torrent.ID > 0 {
 		errUp := uploadForm.ExtractEditInfo(r)
 		if errUp != nil {
 			err["errors"] = append(err["errors"], "Failed to update torrent!")
 		}
-		if (len(err) == 0) {
+		if len(err) == 0 {
 			// update some (but not all!) values
 			torrent.Name = uploadForm.Name
 			torrent.Category = uploadForm.CategoryID

router/template.go

@@ -9,15 +9,18 @@ var TemplateDir = "templates"
 
 var homeTemplate, searchTemplate, faqTemplate, uploadTemplate, viewTemplate, viewRegisterTemplate, viewLoginTemplate, viewRegisterSuccessTemplate, viewVerifySuccessTemplate, viewProfileTemplate, viewProfileEditTemplate, viewUserDeleteTemplate, notFoundTemplate *template.Template
 
+var panelIndex, panelTorrentList, panelUserList, panelCommentList, panelTorrentEd, panelTorrentReportList *template.Template
+
 type templateLoader struct {
 	templ     **template.Template
 	file      string
+	indexFile string
 	name      string
 }
 
 // ReloadTemplates reloads templates on runtime
 func ReloadTemplates() {
-	templs := []templateLoader{
+	pubTempls := []templateLoader{
 		templateLoader{
 			templ: &homeTemplate,
 			name:  "home",
@@ -46,37 +49,37 @@ func ReloadTemplates() {
 		templateLoader{
 			templ: &viewRegisterTemplate,
 			name:  "user_register",
-			file:  "user/register.html",
+			file:  filepath.Join("user", "register.html"),
 		},
 		templateLoader{
 			templ: &viewRegisterSuccessTemplate,
 			name:  "user_register_success",
-			file:  "user/signup_success.html",
+			file:  filepath.Join("user", "signup_success.html"),
 		},
 		templateLoader{
 			templ: &viewVerifySuccessTemplate,
 			name:  "user_verify_success",
-			file:  "user/verify_success.html",
+			file:  filepath.Join("user", "verify_success.html"),
 		},
 		templateLoader{
 			templ: &viewLoginTemplate,
 			name:  "user_login",
-			file:  "user/login.html",
+			file:  filepath.Join("user", "login.html"),
 		},
 		templateLoader{
 			templ: &viewProfileTemplate,
 			name:  "user_profile",
-			file:  "user/profile.html",
+			file:  filepath.Join("user", "profile.html"),
 		},
 		templateLoader{
 			templ: &viewProfileEditTemplate,
 			name:  "user_profile",
-			file:  "user/profile_edit.html",
+			file:  filepath.Join("user", "profile_edit.html"),
 		},
 		templateLoader{
 			templ: &viewUserDeleteTemplate,
 			name:  "user_delete",
-			file:  "user/delete_success.html",
+			file:  filepath.Join("user", "delete_success.html"),
 		},
 		templateLoader{
 			templ: &notFoundTemplate,
@@ -84,10 +87,56 @@ func ReloadTemplates() {
 			file:  "404.html",
 		},
 	}
-	for _, templ := range templs {
-		t := template.Must(template.New(templ.name).Funcs(FuncMap).ParseFiles(filepath.Join(TemplateDir, "index.html"), filepath.Join(TemplateDir, templ.file)))
-		t = template.Must(t.ParseGlob(filepath.Join(TemplateDir, "_*.html")))
+	for idx := range pubTempls {
+		pubTempls[idx].indexFile = filepath.Join(TemplateDir, "index.html")
+	}
 
+	modTempls := []templateLoader{
+		templateLoader{
+			templ: &panelTorrentList,
+			name:  "torrentlist",
+			file:  filepath.Join("admin", "torrentlist.html"),
+		},
+		templateLoader{
+			templ: &panelUserList,
+			name:  "userlist",
+			file:  filepath.Join("admin", "userlist.html"),
+		},
+		templateLoader{
+			templ: &panelCommentList,
+			name:  "commentlist",
+			file:  filepath.Join("admin", "commentlist.html"),
+		},
+		templateLoader{
+			templ: &panelIndex,
+			name:  "indexPanel",
+			file:  filepath.Join("admin", "panelindex.html"),
+		},
+		templateLoader{
+			templ: &panelTorrentEd,
+			name:  "torrent_ed",
+			file:  filepath.Join("admin", "paneltorrentedit.html"),
+		},
+		templateLoader{
+			templ: &panelTorrentReportList,
+			name:  "torrent_report",
+			file:  filepath.Join("admin", "torrent_report.html"),
+		},
+	}
+
+	for idx := range modTempls {
+		modTempls[idx].indexFile = filepath.Join(TemplateDir, "admin_index.html")
+	}
+
+	templs := make([]templateLoader, 0, len(modTempls)+len(pubTempls))
+
+	templs = append(templs, pubTempls...)
+	templs = append(templs, modTempls...)
+
+	for _, templ := range templs {
+		t := template.Must(template.New(templ.name).Funcs(FuncMap).ParseFiles(templ.indexFile, filepath.Join(TemplateDir, templ.file)))
+		t = template.Must(t.ParseGlob(filepath.Join(TemplateDir, "_*.html")))
 		*templ.templ = t
 	}
+
 }

service/scraper/bucket.go

@@ -28,15 +28,42 @@ func (b *Bucket) NewTransaction(swarms []model.Torrent) (t *Transaction) {
 	t = &Transaction{
 		TransactionID: id,
 		bucket:        b,
-		swarms:        swarms,
+		swarms:        make([]model.Torrent, len(swarms)),
 		state:         stateSendID,
 	}
+	copy(t.swarms, swarms)
 	b.transactions[id] = t
 	b.access.Unlock()
 	return
 
 }
 
+func (b *Bucket) ForEachTransaction(v func(uint32, *Transaction)) {
+
+	clone := make(map[uint32]*Transaction)
+
+	b.access.Lock()
+
+	for k := range b.transactions {
+		clone[k] = b.transactions[k]
+	}
+
+	b.access.Unlock()
+
+	for k := range clone {
+		v(k, clone[k])
+	}
+}
+
+func (b *Bucket) Forget(tid uint32) {
+	b.access.Lock()
+	_, ok := b.transactions[tid]
+	if ok {
+		delete(b.transactions, tid)
+	}
+	b.access.Unlock()
+}
+
 func (b *Bucket) VisitTransaction(tid uint32, v func(*Transaction)) {
 	b.access.Lock()
 	t, ok := b.transactions[tid]

service/scraper/scraper.go

@@ -13,6 +13,9 @@ import (
 // MTU yes this is the ipv6 mtu
 const MTU = 1500
 
+// max number of scrapes per packet
+const ScrapesPerPacket = 74
+
 // bittorrent scraper
 type Scraper struct {
 	done             chan int
@@ -21,7 +24,9 @@ type Scraper struct {
 	errQueue         chan error
 	trackers         map[string]*Bucket
 	ticker           *time.Ticker
+	cleanup          *time.Ticker
 	interval         time.Duration
+	PacketsPerSecond uint
 }
 
 func New(conf *config.ScraperConfig) (sc *Scraper, err error) {
@@ -33,7 +38,13 @@ func New(conf *config.ScraperConfig) (sc *Scraper, err error) {
 		trackers:  make(map[string]*Bucket),
 		ticker:    time.NewTicker(time.Second),
 		interval:  time.Second * time.Duration(conf.IntervalSeconds),
+		cleanup:   time.NewTicker(time.Second),
 	}
+
+	if sc.PacketsPerSecond == 0 {
+		sc.PacketsPerSecond = 10
+	}
+
 	for idx := range conf.Trackers {
 		err = sc.AddTracker(&conf.Trackers[idx])
 		if err != nil {
@@ -144,20 +155,37 @@ func (sc *Scraper) RunWorker(pc net.PacketConn) (err error) {
 
 func (sc *Scraper) Run() {
 	for {
-		<-sc.ticker.C
-		sc.Scrape()
+		select {
+		case <-sc.ticker.C:
+			sc.Scrape(sc.PacketsPerSecond)
+			break
+		case <-sc.cleanup.C:
+			sc.removeStale()
+			break
+		}
 	}
 }
 
-func (sc *Scraper) Scrape() {
+func (sc *Scraper) removeStale() {
+
+	for k := range sc.trackers {
+		sc.trackers[k].ForEachTransaction(func(tid uint32, t *Transaction) {
+			if t == nil || t.IsTimedOut() {
+				sc.trackers[k].Forget(tid)
+			}
+		})
+	}
+}
+
+func (sc *Scraper) Scrape(packets uint) {
 	now := time.Now().Add(0 - sc.interval)
 
-	rows, err := db.ORM.Raw("SELECT torrent_id, torrent_hash FROM torrents WHERE last_scrape IS NULL OR last_scrape < ? ORDER BY torrent_id DESC LIMIT 700", now).Rows()
+	rows, err := db.ORM.Raw("SELECT torrent_id, torrent_hash FROM torrents WHERE last_scrape IS NULL OR last_scrape < ? ORDER BY torrent_id DESC LIMIT ?", now, packets*ScrapesPerPacket).Rows()
 	if err == nil {
 		counter := 0
-		var scrape [70]model.Torrent
+		var scrape [ScrapesPerPacket]model.Torrent
 		for rows.Next() {
-			idx := counter % 70
+			idx := counter % ScrapesPerPacket
 			rows.Scan(&scrape[idx].ID, &scrape[idx].Hash)
 			counter++
 			if idx == 0 {

service/scraper/transaction.go

@@ -11,6 +11,9 @@ import (
 	"github.com/ewhal/nyaa/util/log"
 )
 
+// TransactionTimeout 30 second timeout for transactions
+const TransactionTimeout = time.Second * 30
+
 const stateSendID = 0
 const stateRecvID = 1
 const stateTransact = 2
@@ -27,13 +30,12 @@ type Transaction struct {
 	bucket        *Bucket
 	state         uint8
 	swarms        []model.Torrent
+	lastData      time.Time
 }
 
 // Done marks this transaction as done and removes it from parent
 func (t *Transaction) Done() {
-	t.bucket.access.Lock()
-	delete(t.bucket.transactions, t.TransactionID)
-	t.bucket.access.Unlock()
+	t.bucket.Forget(t.TransactionID)
 }
 
 func (t *Transaction) handleScrapeReply(data []byte) {
@@ -95,6 +97,7 @@ func (t *Transaction) SendEvent(to net.Addr) (ev *SendEvent) {
 		binary.BigEndian.PutUint32(ev.Data[12:], t.TransactionID)
 		t.state = stateRecvID
 	}
+	t.lastData = time.Now()
 	return
 }
 
@@ -104,7 +107,7 @@ func (t *Transaction) handleError(msg string) {
 
 // handle data for transaction
 func (t *Transaction) GotData(data []byte) (done bool) {
-
+	t.lastData = time.Now()
 	if len(data) > 4 {
 		cmd := binary.BigEndian.Uint32(data)
 		switch cmd {
@@ -132,3 +135,8 @@ func (t *Transaction) GotData(data []byte) (done bool) {
 	}
 	return
 }
+
+func (t *Transaction) IsTimedOut() bool {
+	return t.lastData.Add(TransactionTimeout).Before(time.Now())
+
+}

util/search/search.go

@@ -75,22 +75,36 @@ func searchByQuery(r *http.Request, pagenum int, countAll bool) (
 	case "1":
 		search.Sort = common.Name
 		orderBy += "torrent_name"
+		break
 	case "2":
 		search.Sort = common.Date
 		orderBy += "date"
+		break
 	case "3":
 		search.Sort = common.Downloads
 		orderBy += "downloads"
+		break
 	case "4":
 		search.Sort = common.Size
 		orderBy += "filesize"
+		break
 	case "5":
+		search.Sort = common.Seeders
 		orderBy += "seeders"
+		search.NotNull += "seeders IS NOT NULL "
+		break
 	case "6":
+		search.Sort = common.Leechers
 		orderBy += "leechers"
+		search.NotNull += "leechers IS NOT NULL "
+		break
 	case "7":
+		search.Sort = common.Completed
 		orderBy += "completed"
+		search.NotNull += "completed IS NOT NULL "
+		break
 	default:
+		search.Sort = common.ID
 		orderBy += "torrent_id"
 	}
 
@@ -129,7 +143,9 @@ func searchByQuery(r *http.Request, pagenum int, countAll bool) (
 			}
 			parameters.Params = append(parameters.Params, strconv.Itoa(int(search.Status)+1))
 		}
-
+		if len(search.NotNull) > 0 {
+			conditions = append(conditions, search.NotNull)
+		}
 		searchQuerySplit := strings.Fields(search.Query)
 		for i, word := range searchQuerySplit {
 			firstRune, _ := utf8.DecodeRuneInString(word)