fix link and h2 text lost

wedojava · Sep 30, 2020 · 4852a29 · 4852a29
1 parent 36fb1b3
commit 4852a29
Show file tree

Hide file tree

Showing 2 changed files with 6 additions and 5 deletions.
diff --git a/internal/fetcher/sites/cna/cna.go b/internal/fetcher/sites/cna/cna.go
@@ -134,7 +134,7 @@ func cna(p *Post) (string, error) {
 		return "", errors.New("[-] There is no element class is paragraph` from: " + p.URL.String())
 	}
 	n := nodes[0]
-	plist := htmldoc.ElementsByTag(n, "p")
+	plist := htmldoc.ElementsByTag(n, "h2", "p")
 	for _, v := range plist {
 		if v.FirstChild != nil {
 			body += v.FirstChild.Data + "  \n"
@@ -143,8 +143,9 @@ func cna(p *Post) (string, error) {
 
 	body = strings.ReplaceAll(body, "「", "“")
 	body = strings.ReplaceAll(body, "」", "”")
+	body = strings.ReplaceAll(body, "</a>", "")
 
-	re := regexp.MustCompile(`<a.*?</a>`)
+	re := regexp.MustCompile(`<a.*?>`)
 	body = re.ReplaceAllString(body, "")
 	re = regexp.MustCompile(`<iframe.*?</iframe>`)
 	body = re.ReplaceAllString(body, "")

diff --git a/internal/fetcher/sites/cna/cna_test.go b/internal/fetcher/sites/cna/cna_test.go
@@ -10,7 +10,7 @@ import (
 	"github.com/wedojava/fetcher/internal/htmldoc"
 )
 
-var p = PostFactory("https://www.cna.com.tw/news/firstnews/202009295001.aspx")
+var p = PostFactory("https://www.cna.com.tw/news/aopl/202009300058.aspx")
 
 func PostFactory(rawurl string) *Post {
 	url, err := url.Parse(rawurl)
@@ -32,7 +32,7 @@ func TestSetDate(t *testing.T) {
 	if err := setDate(p); err != nil {
 		t.Errorf("test SetPost err: %v", doc)
 	}
-	want := "2020-09-29T11:49:00+08:00"
+	want := "2020-09-30T10:54:00+08:00"
 	if p.Date != want {
 		t.Errorf("\ngot: %v\nwant: %v", p.Date, want)
 	}
@@ -47,7 +47,7 @@ func TestSetTitle(t *testing.T) {
 	if err := setTitle(p); err != nil {
 		t.Errorf("test SetPost err: %v", err)
 	}
-	want := "早安世界》安心旅遊補助續辦至10月底 中秋雙十連假可用 | 生活 | 重點新聞"
+	want := "被爆10年沒繳稅 川普：避稅計畫展現我的才智 | 國際"
 	if p.Title != want {
 		t.Errorf("\ngot: %v\nwant: %v", p.Title, want)
 	}