brokenlinks: mark the link in queue as seen with status code 0

This is to fix double URL being pushed to queue.

1 files changed, 11 insertions, 10 deletions

diff --git a/brokenlinks/worker.go b/brokenlinks/worker.go
index 4a9a1d9..09c8b12 100644
--- a/brokenlinks/worker.go
+++ b/brokenlinks/worker.go
@@ -178,14 +178,14 @@ func (wrk *worker) scanPastResult() (result *Result, err error) {
 //   - Otherwise push it to queue.
 func (wrk *worker) processResult(resultq map[string]jarink.Link) {
 	var seen bool
-	for _, linkq := range resultq {
-		if linkq.StatusCode != 0 {
-			wrk.seen(linkq)
+	for _, result := range resultq {
+		if result.StatusCode != 0 {
+			wrk.seen(result)
 			continue
 		}
 
-		if linkq.IsExternal {
-			var scannedLink = wrk.cache.Get(linkq.Url)
+		if result.IsExternal {
+			var scannedLink = wrk.cache.Get(result.Url)
 			if scannedLink != nil {
 				// The external link has been scanned
 				// previously.
@@ -193,17 +193,18 @@ func (wrk *worker) processResult(resultq map[string]jarink.Link) {
 			}
 		}
 
-		linkq.StatusCode, seen = wrk.seenLink[linkq.Url]
+		result.StatusCode, seen = wrk.seenLink[result.Url]
 		if seen {
-			if linkq.StatusCode >= http.StatusBadRequest {
+			if result.StatusCode >= http.StatusBadRequest {
 				// Different pages may have the same broken
 				// link.
-				wrk.markAsBroken(linkq)
+				wrk.markAsBroken(result)
 			}
 			continue
 		}
-		wrk.queue = append(wrk.queue, linkq)
-		wrk.log.Printf(`queue %d: %s`, len(wrk.queue), linkq.Url)
+		wrk.queue = append(wrk.queue, result)
+		wrk.seenLink[result.Url] = result.StatusCode
+		wrk.log.Printf(`queue %d: %s`, len(wrk.queue), result.Url)
 	}
 }