func MimeDetector(sf types.SetupFunction) {
in, out := sf.SetName("MimeDetector").AsFilter("UrlParser").Build()
defer close(out)
buffer := make([]byte, 512)
for urlBM := range in {
if sf.Cancelled() {
continue
}
url := ToString(urlBM)
body := Download(url)
if body == nil {
continue
}
func(body io.ReadCloser) {
defer body.Close()
n, err := body.Read(buffer)
if err == nil || err == io.EOF {
encoded := http.DetectContentType(buffer[:n]) + "->" + url
out <- NewStringMarshaler(encoded)
}
}(body)
}
}
func UrlParser(sf types.SetupFunction) {
in, out := sf.SetName("UrlParser").AsFilter("UrlProducer").Build()
defer close(out)
for urlBM := range in {
if sf.Cancelled() {
continue
}
urlStr := ToString(urlBM)
if path.Ext(urlStr) == "" && urlStr[len(urlStr)-1] != '/' {
urlStr += "/"
}
url, err := urlParser.Parse(urlStr)
if err != nil {
continue
}
body := Download(url.String())
for link := range fetchLinks(body) {
joinedLink, err := url.Parse(link)
if err != nil {
continue
}
out <- NewStringMarshaler(joinedLink.String())
}
}
}
func pathProducer(sf types.SetupFunction, argv []string) {
out := sf.SetName("PathProducer").AsProducer().Build()
defer close(out)
for _, path := range argv {
out <- NewStringMarshaler(path)
}
}
func PathValidator(sf types.SetupFunction, parent string) {
in, out := sf.SetName("PathValidator").AsFilter(parent).Build()
defer close(out)
for path := range in {
if _, err := os.Stat(ToString(path)); err == nil {
out <- path
}
}
}
func urlLooper(sf types.SetupFunction, feeder chan<- interface{}, done types.AtomicBool) {
in := sf.AsConsumer("MimeSplitterHtml").Build()
for url := range in {
if !done.Get() {
u := ToString(url)
feeder <- u
}
}
}
func urlFeeder(sf types.SetupFunction, feeder <-chan interface{}, done types.AtomicBool) {
out := sf.SetName("UrlFeeder").AsProducer().Build()
defer close(out)
for url := range feeder {
if !done.Get() {
out <- NewStringMarshaler(url.(string))
}
}
}
func wordPrinter(sf types.SetupFunction, done chan struct{}) {
defer close(done)
in := sf.AsConsumer("github.com/apoydence/hydra/examples/wordCount.FinalWordCounter").Build()
for wordMap := range in {
for k, v := range ToMap(wordMap) {
println(k, v)
}
}
}
func WordCounter(sf types.SetupFunction) {
in, out := sf.AsFilter("github.com/apoydence/hydra/examples/wordCount.SymbolRemover").Build()
defer close(out)
m := make(map[string]uint32)
for word := range in {
incMap(ToString(word), 1, m)
}
out <- NewWordCountMarshaler(m)
}
func FinalWordCounter(sf types.SetupFunction) {
in, out := sf.AsFilter("github.com/apoydence/hydra/examples/wordCount.WordCounter").Build()
defer close(out)
m := make(map[string]uint32)
for wordMap := range in {
for k, v := range ToMap(wordMap) {
incMap(k, v, m)
}
}
out <- NewWordCountMarshaler(m)
}
func MimeSplitterHtml(sf types.SetupFunction) {
in, out := sf.SetName("MimeSplitterHtml").AsFilter("MimeDetector").Build()
defer close(out)
for urlBM := range in {
if sf.Cancelled() {
continue
}
url := ToString(urlBM)
mime, u := decodeMimeUrl(url)
if strings.Contains(mime, "html") {
out <- NewStringMarshaler(u)
}
}
}
func SymbolRemover(sf types.SetupFunction) {
in, out := sf.AsFilter("github.com/apoydence/hydra/examples/wordCount.WordExtractor").Build()
defer close(out)
for word := range in {
str := make([]byte, 0)
bytes := []byte(strings.ToLower(ToString(word)))
for _, x := range bytes {
if (x >= 0x30 && x <= 0x39) || (x >= 0x61 && x <= 0x7a) {
str = append(str, x)
}
}
if len(str) > 0 {
out <- NewStringMarshaler(string(str))
}
}
}
func WordExtractor(sf types.SetupFunction) {
in, out := sf.AsFilter("PathValidator").Build()
defer close(out)
for path := range in {
p := ToString(path)
f, err := os.Open(p)
if err == nil {
scanner := bufio.NewScanner(f)
scanner.Split(bufio.ScanWords)
for scanner.Scan() {
out <- NewStringMarshaler(scanner.Text())
}
}
}
}
func UrlProducer(sf types.SetupFunction, parent string) {
in, out := sf.SetName("UrlProducer").AsFilter(parent).Build()
defer close(out)
visitedUrls := make(map[string]interface{})
for urlBm := range in {
if sf.Cancelled() {
continue
}
url := ToString(urlBm)
if _, visited := visitedUrls[url]; !visited {
visitedUrls[url] = nil
out <- NewStringMarshaler(url)
}
}
}
func textDownloader(sf types.SetupFunction, download string, mb int, done chan struct{}, closer func()) {
in := sf.AsConsumer("MimeSplitterText").Build()
var totalSize int64 = 0
for url := range in {
if totalSize >= int64(mb)*1024*1024 {
closer()
continue
}
u := ToString(url)
println("download", u, path.Base(u))
totalSize += saveToFile(Download(u), path.Join(download, path.Base(u)))
println("Size", totalSize)
}
close(done)
}
