aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorTucker Evans <tuckerevans24@gmail.com>2017-12-02 16:38:30 -0500
committerTucker Evans <tuckerevans24@gmail.com>2019-02-17 20:33:00 -0500
commitac839180a1a3c8477989d887734798da146d7b57 (patch)
tree383c63874892ed63126e5905f49c32e0096e8b68
parentde031c361413fccae6e9c516ca9866825d350253 (diff)
CSC2621/assignments/webCrawler2: testing merge --theirs
-rw-r--r--webCrawler2/crawler.go9
1 files changed, 4 insertions, 5 deletions
diff --git a/webCrawler2/crawler.go b/webCrawler2/crawler.go
index 1c7bc70..5c4dba6 100644
--- a/webCrawler2/crawler.go
+++ b/webCrawler2/crawler.go
@@ -27,7 +27,7 @@ func validLink(s string) bool {
//return (strings.HasSuffix(s, ".html") || strings.HasSuffix(s, "/") || strings.HasSuffix(s, "\\"))
}
-func addLinks(doc *goquery.Document, jobs chan link, current link, depth int) {
+func addLinks(doc *goquery.Document, jobs chan link, current link, depth int, worker_id int) {
doc.Find("body a").Each(func(index int, item *goquery.Selection) {
link_s, _ := item.Attr("href")
@@ -71,7 +71,6 @@ func worker(done chan bool, jobs chan link, depth int, id int, total uint64) {
case j := <-jobs:
if j.depth < depth {
doc, err := goquery.NewDocument(j.u.String())
- docs <- doc
if err != nil {
log.Print("Error Reading Document: " + j.u.String() + err.Error())
break
@@ -83,6 +82,7 @@ func worker(done chan bool, jobs chan link, depth int, id int, total uint64) {
addLinks(doc, jobs, j, j.depth, id)
}
case <-time.After(time.Second * 10):
+ fmt.Printf("Worker %d done\n", id)
done <- true
return
}
@@ -129,8 +129,7 @@ func main() {
t = uint64(b)
w, _ = strconv.Atoi(os.Args[4])
- links := make(chan link, 1024*1024)
- docs := make(chan *goquery.Document, 100)
+ jobs := make(chan link, 1024*1024)
done := make(chan bool)
u, err := url.Parse(os.Args[1])
@@ -141,7 +140,7 @@ func main() {
if !u.IsAbs() {
panic("Cannot start with relative url")
}
- links <- link{u, 0}
+ jobs <- link{u, 0}
//send first job