func main() {
defer func() {
tmpFn := villa.Path("/tmp/gddo")
if err := tmpFn.RemoveAll(); err != nil {
log.Printf("Delete %v failed: %v", tmpFn, err)
}
}()
singlePackge := ""
singleETag := ""
flag.StringVar(&singlePackge, "pkg", singlePackge, "Crawling single package")
flag.StringVar(&singleETag, "etag", singleETag, "ETag for single package crawling")
flag.Parse()
httpClient := gcse.GenHttpClient("")
if singlePackge != "" {
log.Printf("Crawling single package %s ...", singlePackge)
p, err := gcse.CrawlPackage(httpClient, singlePackge, singleETag)
if err != nil {
fmtp.Printfln("Crawling package %s failured: %v", singlePackge, err)
} else {
fmtp.Printfln("Package %s: %+v", singlePackge, p)
}
return
}
log.Println("crawler started...")
// Load CrawlerDB
cDB = gcse.LoadCrawlerDB()
fpDataRoot := sophie.FsPath{
Fs: sophie.LocalFS,
Path: gcse.DataRoot.S(),
}
fpDocs := fpDataRoot.Join(gcse.FnDocs)
if err := loadAllDocsPkgs(kv.DirInput(fpDocs)); err != nil {
log.Fatalf("loadAllDocsPkgs: %v", err)
}
log.Printf("%d docs loaded!", len(allDocsPkgs))
AppStopTime = time.Now().Add(gcse.CrawlerDuePerRun)
//pathToCrawl := gcse.DataRoot.Join(gcse.FnToCrawl)
fpCrawler := fpDataRoot.Join(gcse.FnCrawlerDB)
fpToCrawl := fpDataRoot.Join(gcse.FnToCrawl)
fpNewDocs := fpCrawler.Join(gcse.FnNewDocs)
fpNewDocs.Remove()
pkgEnd := make(chan error, 1)
go crawlPackages(httpClient, fpToCrawl.Join(gcse.FnPackage), fpNewDocs,
pkgEnd)
psnEnd := make(chan error, 1)
go crawlPersons(httpClient, fpToCrawl.Join(gcse.FnPerson), psnEnd)
errPkg, errPsn := <-pkgEnd, <-psnEnd
if errPkg != nil || errPsn != nil {
log.Fatalf("Some job may failed, package: %v, person: %v",
errPkg, errPsn)
}
if err := processImports(); err != nil {
log.Printf("processImports failed: %v", err)
}
syncDatabases()
log.Println("crawler stopped...")
}
// OnlyMapper.Map
func (pc *PackageCrawler) Map(key, val sophie.SophieWriter, c []sophie.Collector) error {
if time.Now().After(AppStopTime) {
log.Printf("[Part %d] Timeout(key = %v), PackageCrawler returns EOM",
pc.part, key)
return mr.EOM
}
pkg := string(*key.(*sophie.RawString))
ent := val.(*gcse.CrawlingEntry)
if ent.Version < gcse.CrawlerVersion {
// if gcse.CrawlerVersion is larger than Version, Etag is ignored.
ent.Etag = ""
}
log.Printf("[Part %d] Crawling package %v with etag %s\n", pc.part, pkg, ent.Etag)
p, flds, err := gcse.CrawlPackage(pc.httpClient, pkg, ent.Etag)
for _, fld := range flds {
if spider.LikeGoSubFolder(fld.Name) {
appendNewPackage(pkg+"/"+fld.Path, "parent")
}
}
site, path := utils.SplitPackage(pkg)
if err != nil && errorsp.Cause(err) != gcse.ErrPackageNotModifed {
log.Printf("[Part %d] Crawling pkg %s failed: %v", pc.part, pkg, err)
if gcse.IsBadPackage(err) {
utils.LogError(store.AppendPackageEvent(site, path, "", time.Now(), sppb.HistoryEvent_Action_Invalid), "AppendPackageEvent %v %v failed", site, path)
bi.AddValue(bi.Sum, "crawler.package.wrong-package", 1)
// a wrong path
nda := gcse.NewDocAction{
Action: gcse.NDA_DEL,
}
c[0].Collect(sophie.RawString(pkg), &nda)
cDB.PackageDB.Delete(pkg)
log.Printf("[Part %d] Remove wrong package %s", pc.part, pkg)
} else {
utils.LogError(store.AppendPackageEvent(site, path, "", time.Now(), sppb.HistoryEvent_Action_Failed), "AppendPackageEvent %v %v failed", site, path)
bi.Inc("crawler.package.failed")
if strings.HasPrefix(pkg, "github.com/") {
bi.Inc("crawler.package.failed.github")
}
pc.failCount++
cDB.SchedulePackage(pkg, time.Now().Add(12*time.Hour), ent.Etag)
if pc.failCount >= 10 || strings.Contains(err.Error(), "403") {
durToSleep := 10 * time.Minute
if time.Now().Add(durToSleep).After(AppStopTime) {
log.Printf("[Part %d] Timeout(key = %v), PackageCrawler returns EOM",
pc.part, key)
return mr.EOM
}
log.Printf("[Part %d] Last ten crawling packages failed, sleep for a while...(current: %s)",
pc.part, pkg)
time.Sleep(durToSleep)
pc.failCount = 0
}
}
return nil
}
utils.LogError(store.AppendPackageEvent(site, path, "", time.Now(), sppb.HistoryEvent_Action_Success), "AppendPackageEvent %v %v failed", site, path)
pc.failCount = 0
if errorsp.Cause(err) == gcse.ErrPackageNotModifed {
// TODO crawling stars for unchanged project
log.Printf("[Part %d] Package %s unchanged!", pc.part, pkg)
schedulePackageNextCrawl(pkg, ent.Etag)
bi.AddValue(bi.Sum, "crawler.package.not-modified", 1)
return nil
}
bi.AddValue(bi.Sum, "crawler.package.success", 1)
if strings.HasPrefix(pkg, "github.com/") {
bi.AddValue(bi.Sum, "crawler.package.success.github", 1)
}
log.Printf("[Part %d] Crawled package %s success!", pc.part, pkg)
var pkgInfo *stpb.PackageInfo
if err := store.UpdatePackage(site, path, func(pi *stpb.PackageInfo) error {
fillPackageInfo(p, pi)
pkgInfo = pi
return nil
}); err != nil {
log.Printf("UpdatePackage %v %v failed: %v", site, path, err)
}
saveRelatedInfo(pkgInfo)
nda := gcse.NewDocAction{
Action: gcse.NDA_UPDATE,
DocInfo: packageToDoc(p),
}
c[0].Collect(sophie.RawString(pkg), &nda)
log.Printf("[Part %d] Package %s saved!", pc.part, pkg)
if !strings.HasPrefix(pkg, "github.com/") {
// github.com throttling is done within the GithubSpider.
time.Sleep(10 * time.Second)
}
return nil
}
func crawlEntriesLoop() {
httpClient := gcse.GenHttpClient("")
for time.Now().Before(AppStopTime) {
checkImports()
if gcse.CrawlByGodocApi {
processGodoc(httpClient)
}
didSomething := false
var wg sync.WaitGroup
pkgGroups := listPackagesByHost(5, 50)
if len(pkgGroups) > 0 {
didSomething = true
log.Printf("Crawling packages of %d groups", len(pkgGroups))
wg.Add(len(pkgGroups))
for host, ents := range pkgGroups {
go func(host string, ents []EntryInfo) {
failCount := 0
for _, ent := range ents {
if time.Now().After(AppStopTime) {
break
}
runtime.GC()
p, err := gcse.CrawlPackage(httpClient, ent.ID, ent.Etag)
if err != nil && err != gcse.ErrPackageNotModifed {
log.Printf("Crawling pkg %s failed: %v", ent.ID, err)
if gcse.IsBadPackage(err) {
// a wrong path
deletePackage(ent.ID)
log.Printf("Remove wrong package %s", ent.ID)
} else {
failCount++
schedulePackage(ent.ID, time.Now().Add(
12*time.Hour), ent.Etag)
if failCount >= 10 {
durToSleep := 10 * time.Minute
if time.Now().Add(durToSleep).After(AppStopTime) {
break
}
log.Printf("Last ten crawling %s packages failed, sleep for a while...",
host)
time.Sleep(durToSleep)
failCount = 0
}
}
continue
}
failCount = 0
if err == gcse.ErrPackageNotModifed {
log.Printf("Package %s unchanged!", ent.ID)
schedulePackageNextCrawl(ent.ID, ent.Etag)
continue
}
log.Printf("Crawled package %s success!", ent.ID)
pushPackage(p)
log.Printf("Package %s saved!", ent.ID)
}
wg.Done()
}(host, ents)
}
}
personGroups := listPersonsByHost(5, 100)
if len(personGroups) > 0 {
didSomething = true
log.Printf("Crawling persons of %d groups", len(personGroups))
wg.Add(len(personGroups))
for host, ents := range personGroups {
go func(host string, ents []EntryInfo) {
failCount := 0
for _, ent := range ents {
if time.Now().After(AppStopTime) {
break
}
p, err := gcse.CrawlPerson(httpClient, ent.ID)
if err != nil {
failCount++
log.Printf("Crawling person %s failed: %v", ent.ID, err)
schedulePerson(ent.ID, time.Now().Add(12*time.Hour))
if failCount >= 10 {
durToSleep := 10 * time.Minute
if time.Now().Add(durToSleep).After(AppStopTime) {
break
}
log.Printf("Last ten crawling %s persons failed, sleep for a while...",
host)
time.Sleep(durToSleep)
failCount = 0
}
continue
}
log.Printf("Crawled person %s success!", ent.ID)
pushPerson(p)
log.Printf("Push person %s success", ent.ID)
failCount = 0
}
wg.Done()
}(host, ents)
}
}
wg.Wait()
syncDatabases()
if gcse.CrawlGithubUpdate {
if touchByGithubUpdates() {
didSomething = true
}
}
if !didSomething {
log.Printf("Nothing to crawl sleep for a while...")
time.Sleep(2 * time.Minute)
}
}
}
// OnlyMapper.Map
func (pc *PackageCrawler) Map(key, val sophie.SophieWriter,
c []sophie.Collector) error {
if time.Now().After(AppStopTime) {
log.Printf("[Part %d] Timeout(key = %v), PackageCrawler returns EOM",
pc.part, key)
return mr.EOM
}
pkg := string(*key.(*sophie.RawString))
ent := val.(*gcse.CrawlingEntry)
if ent.Version < gcse.CrawlerVersion {
// if gcse.CrawlerVersion is larger than Version, Etag is ignored.
ent.Etag = ""
}
log.Printf("[Part %d] Crawling package %v with etag %s\n", pc.part, pkg, ent.Etag)
p, err := gcse.CrawlPackage(pc.httpClient, pkg, ent.Etag)
_ = p
if err != nil && err != gcse.ErrPackageNotModifed {
log.Printf("[Part %d] Crawling pkg %s failed: %v", pc.part, pkg, err)
if gcse.IsBadPackage(err) {
// a wrong path
nda := gcse.NewDocAction{
Action: gcse.NDA_DEL,
}
c[0].Collect(sophie.RawString(pkg), &nda)
cDB.PackageDB.Delete(pkg)
log.Printf("[Part %d] Remove wrong package %s", pc.part, pkg)
} else {
pc.failCount++
cDB.SchedulePackage(pkg, time.Now().Add(12*time.Hour), ent.Etag)
if pc.failCount >= 10 || strings.Contains(err.Error(), "403") {
durToSleep := 10 * time.Minute
if time.Now().Add(durToSleep).After(AppStopTime) {
log.Printf("[Part %d] Timeout(key = %v), PackageCrawler returns EOM",
pc.part, key)
return mr.EOM
}
log.Printf("[Part %d] Last ten crawling packages failed, sleep for a while...(current: %s)",
pc.part, pkg)
time.Sleep(durToSleep)
pc.failCount = 0
}
}
return nil
}
pc.failCount = 0
if err == gcse.ErrPackageNotModifed {
// TODO crawling stars for unchanged project
log.Printf("[Part %d] Package %s unchanged!", pc.part, pkg)
schedulePackageNextCrawl(pkg, ent.Etag)
return nil
}
log.Printf("[Part %d] Crawled package %s success!", pc.part, pkg)
nda := gcse.NewDocAction{
Action: gcse.NDA_UPDATE,
DocInfo: packageToDoc(p),
}
c[0].Collect(sophie.RawString(pkg), &nda)
log.Printf("[Part %d] Package %s saved!", pc.part, pkg)
time.Sleep(10 * time.Second)
return nil
}
func main() {
runtime.GOMAXPROCS(2)
log.Printf("Using personal: %v", configs.CrawlerGithubPersonal)
gcse.GithubSpider = github.NewSpiderWithToken(configs.CrawlerGithubPersonal)
if db, err := bh.Open(configs.DataRoot.Join("filecache.bolt").S(), 0644, nil); err == nil {
log.Print("Using file cache!")
gcse.GithubSpider.FileCache = spider.BoltFileCache{
DB: db,
IncCounter: bi.Inc,
}
} else {
log.Printf("Open file cache failed: %v", err)
}
cleanTempDir()
defer cleanTempDir()
singlePackage := flag.String("pkg", "", "Crawling a single package")
singleETag := flag.String("etag", "", "ETag for the single package crawling")
singlePerson := flag.String("person", "", "Crawling a single person")
flag.Parse()
httpClient := gcse.GenHttpClient("")
if *singlePerson != "" {
log.Printf("Crawling single person %s ...", *singlePerson)
p, err := gcse.CrawlPerson(httpClient, *singlePerson)
if err != nil {
fmtp.Printfln("Crawling person %s failed: %v", *singlePerson, err)
} else {
fmtp.Printfln("Person %s: %+v", *singlePerson, p)
}
}
if *singlePackage != "" {
log.Printf("Crawling single package %s ...", *singlePackage)
p, flds, err := gcse.CrawlPackage(httpClient, *singlePackage, *singleETag)
if err != nil {
fmtp.Printfln("Crawling package %s failed: %v, folders: %v", *singlePackage, err, flds)
} else {
fmtp.Printfln("Package %s: %+v, folders: %v", *singlePackage, p, flds)
}
}
if *singlePackage != "" || *singlePerson != "" {
return
}
log.Println("crawler started...")
// Load CrawlerDB
cDB = gcse.LoadCrawlerDB()
fpDataRoot := sophie.FsPath{
Fs: sophie.LocalFS,
Path: configs.DataRoot.S(),
}
fpDocs := fpDataRoot.Join(configs.FnDocs)
if err := loadAllDocsPkgs(kv.DirInput(fpDocs)); err != nil {
log.Fatalf("loadAllDocsPkgs: %v", err)
}
log.Printf("%d docs loaded!", len(allDocsPkgs))
AppStopTime = time.Now().Add(configs.CrawlerDuePerRun)
//pathToCrawl := gcse.DataRoot.Join(gcse.FnToCrawl)
fpCrawler := fpDataRoot.Join(configs.FnCrawlerDB)
fpToCrawl := fpDataRoot.Join(configs.FnToCrawl)
fpNewDocs := fpCrawler.Join(configs.FnNewDocs)
fpNewDocs.Remove()
if err := processImports(); err != nil {
log.Printf("processImports failed: %v", err)
}
pkgEnd := make(chan error, 1)
go crawlPackages(httpClient, fpToCrawl.Join(configs.FnPackage), fpNewDocs, pkgEnd)
psnEnd := make(chan error, 1)
go crawlPersons(httpClient, fpToCrawl.Join(configs.FnPerson), psnEnd)
errPkg, errPsn := <-pkgEnd, <-psnEnd
bi.Flush()
bi.Process()
syncDatabases()
if errPkg != nil || errPsn != nil {
log.Fatalf("Some job may failed, package: %v, person: %v", errPkg, errPsn)
}
log.Println("crawler stopped...")
}