func FetchAllOutgoingLinks(uname string) {
getResp, err := http.Get("http://localhost:8080/JSoupRestAPIService/jsoup-api/jsoupapi/" + uname)
if err != nil {
fmt.Printf("Error in jsoup parsing %s \n", err.Error())
} else {
rb, err := ioutil.ReadAll(getResp.Body)
//fmt.Printf("%s \n",rb)
// Check for error
if err != nil {
fmt.Printf("Error in read parsing \n")
} else {
var outlinks []LinkStructure
json.Unmarshal(rb, &outlinks)
var fetchedurls map[string]bool = make(map[string]bool)
for i := 0; i < len(outlinks); i++ {
//fmt.Printf("\nurl = %s\n",outlinks[0].Link)
str := strings.TrimSpace(outlinks[i].Link)
obj := priorityQueue.NewURL(str, 0, false)
obj.Canonicalize()
strname := obj.Geturlname()
count, ok := seenURLS[strname]
if !ok {
count, ok := queuedURLS[strname]
if !ok {
_, ok := fetchedurls[strname]
if !ok && Isurlok(strname) && IsHTMLDoc(strname) {
heap.Push(pq, obj)
//fmt.Printf("\nadd %s to heap\n",strname)
fetchedurls[strname] = true
queuedURLS[strname] = 1
}
} else {
queuedURLS[strname] = count + 1
}
} else {
seenURLS[strname] = count + 1
}
}
f, err := os.OpenFile("linkGraph.txt", os.O_RDWR|os.O_APPEND|os.O_CREATE, 0660)
if err != nil {
fmt.Printf("Error writing to file %s \n", err)
} else {
w := bufio.NewWriter(f)
w.WriteString(uname + "\t")
for key, _ := range fetchedurls {
w.WriteString(key + "\t")
//fmt.Printf("\nadd %s to heap",key)
}
w.WriteString("\n")
w.Flush()
}
}
}
}
func main() {
fmt.Println("Hello World!")
heap.Init(pq)
file, err := os.Open("seedurls")
if err != nil {
log.Fatal(err)
}
defer file.Close()
scanner := bufio.NewScanner(file)
for scanner.Scan() {
obj := priorityQueue.NewURL(strings.TrimSpace(scanner.Text()), 0, true)
obj.Canonicalize()
queuedURLS[obj.Geturlname()] = 0
heap.Push(pq, obj)
}
for pq.Len() > 0 && len(seenURLS) <= 10 {
var temp map[*priorityQueue.URL]bool = make(map[*priorityQueue.URL]bool)
currUrl := heap.Pop(pq).(*priorityQueue.URL)
var name string = currUrl.Geturlname()
delete(queuedURLS, name)
//fmt.Printf("considering for url %s\n",name)
hostname, _ := url.Parse(name)
_, ok := domainTimes[hostname.Host]
for ok && pq.Len() > 0 {
//fmt.Printf("did not reach here\n")
temp[currUrl] = true
currUrl := heap.Pop(pq).(*priorityQueue.URL)
var name string = currUrl.Geturlname()
delete(queuedURLS, name)
hostname, _ := url.Parse(name)
_, ok1 := domainTimes[hostname.Host]
ok = ok1
}
if !ok {
//fmt.Printf("not found in timestamps \n")
ProcessURL(currUrl)
seenURLS[currUrl.Geturlname()] = 0
domainTimes[hostname.Host] = time.Now().Unix()
} else {
temp[currUrl] = true
time.Sleep(time.Second * 2)
}
for key, _ := range temp {
queuedURLS[key.Geturlname()] = 0
heap.Push(pq, key)
}
removeUnallowedDomains()
}
}
