brokenlinks: check for redirect during scan

If the request redirected, use the "Location" value in the response
header as the parent URL instead of from the original link in queue.

2 files changed, 12 insertions, 2 deletions

diff --git a/brokenlinks/worker.go b/brokenlinks/worker.go
index 7683730..06bdcc6 100644
--- a/brokenlinks/worker.go
+++ b/brokenlinks/worker.go
@@ -282,11 +282,17 @@ func (wrk *worker) scan(linkq jarink.Link) (resultq map[string]jarink.Link) {
 
 	var parentUrl *url.URL
 
-	parentUrl, err = url.Parse(linkq.Url)
+	// Check and get the redirect location or use the original URL.
+	location := httpResp.Header.Get(`Location`)
+	if location == `` {
+		location = linkq.Url
+	}
+
+	parentUrl, err = url.Parse(location)
 	if err != nil {
 		linkq.StatusCode = StatusBadLink
 		linkq.ErrScan = err
-		resultq[linkq.Url] = linkq
+		resultq[location] = linkq
 		return resultq
 	}
 
diff --git a/brokenlinks/worker_test.go b/brokenlinks/worker_test.go
index 2867338..122221e 100644
--- a/brokenlinks/worker_test.go
+++ b/brokenlinks/worker_test.go
@@ -35,6 +35,10 @@ func TestGenURLRelative(t *testing.T) {
 		parentURL: `https://domain/a/b/page.HTML`,
 		relPath:   `c`,
 		expURL:    `https://domain/a/b/c`,
+	}, {
+		parentURL: `https://domain/a/b/page.HTML`,
+		relPath:   `../c.html`,
+		expURL:    `https://domain/a/c.html`,
 	}}
 	for _, tc := range listCase {
 		parentURL, _ := url.Parse(tc.parentURL)