Refactor (#237)

* Refactor * Refactor * Add unit tests * Refactor * Add test
raviqqe · Aug 1, 2022 · 82068a2 · 82068a2
1 parent ff097ab
commit 82068a2
Show file tree

Hide file tree

Showing 2 changed files with 35 additions and 7 deletions.
diff --git a/link_finder.go b/link_finder.go
@@ -51,19 +51,20 @@ func (f linkFinder) Find(n *html.Node, base *url.URL) map[string]error {
 			for _, s := range ss {
 				s := strings.TrimSpace(s)
 
+				if s == "" {
+					continue
+				}
+
 				u, err := url.Parse(s)
 				if err != nil {
 					ls[s] = err
 					continue
 				}
-				fullurl := base.ResolveReference(u).String()
 
-				if s == "" || f.isLinkExcluded(fullurl) || len(f.includedPatterns) > 0 && !f.isLinkIncluded(fullurl) {
-					continue
-				}
+				s = base.ResolveReference(u).String()
 
-				if _, ok := validSchemes[u.Scheme]; ok {
-					ls[fullurl] = nil
+				if _, ok := validSchemes[u.Scheme]; ok && !f.isLinkExcluded(s) && f.isLinkIncluded(s) {
+					ls[s] = nil
 				}
 			}
 		}
@@ -92,7 +93,7 @@ func (f linkFinder) isLinkExcluded(u string) bool {
 }
 
 func (f linkFinder) isLinkIncluded(u string) bool {
-	return f.matches(u, f.includedPatterns)
+	return len(f.includedPatterns) == 0 || f.matches(u, f.includedPatterns)
 }
 
 func (f linkFinder) matches(u string, rs []*regexp.Regexp) bool {

diff --git a/link_finder_test.go b/link_finder_test.go
@@ -19,6 +19,7 @@ func TestLinkFinderFindLinks(t *testing.T) {
 		linkCount int
 	}{
 		{``, 0},
+		{`<a href="" />`, 0},
 		{`<a href="/" />`, 1},
 		{`<a href="/foo" />`, 1},
 		// TODO: Test <frame> tag.
@@ -170,6 +171,32 @@ func TestLinkFinderIsLinkIncluded(t *testing.T) {
 	}
 }
 
+func TestLinkFinderExcludeEntireUrl(t *testing.T) {
+	b, err := url.Parse("http://foo.com")
+	assert.Nil(t, err)
+
+	n, err := html.Parse(strings.NewReader(htmlWithBody(`<a href="/bar" />`)))
+	assert.Nil(t, err)
+
+	rs, err := compileRegexps([]string{"foo"})
+	assert.Nil(t, err)
+
+	assert.Equal(t, map[string]error{}, newLinkFinder(rs, nil).Find(n, b))
+}
+
+func TestLinkFinderIncludeEntireUrl(t *testing.T) {
+	b, err := url.Parse("http://foo.com")
+	assert.Nil(t, err)
+
+	n, err := html.Parse(strings.NewReader(htmlWithBody(`<a href="/bar" />`)))
+	assert.Nil(t, err)
+
+	rs, err := compileRegexps([]string{"foo"})
+	assert.Nil(t, err)
+
+	assert.Equal(t, map[string]error{"http://foo.com/bar": nil}, newLinkFinder(nil, rs).Find(n, b))
+}
+
 func TestLinkFinderFindLinkInSrcSet(t *testing.T) {
 	b, err := url.Parse("http://foo.com")
 	assert.Nil(t, err)