p209327/nlnieuws
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

HLN, Parool: fix relatief xpath

Browse Source
This commit is contained in:
Peter Kleiweg
2026-05-23 14:43:07 +02:00
parent 81cc653ee7
commit 4b56c0cd70
2 changed files with 13 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
HLN/cmd/hln/hln.go
View File

@@ -185,7 +185,7 @@ func doArticle(filename string, url string, timestamp time.Time, needUpdate bool
article := articles[0]
tags := make([]string, 0)
ell, err := article.Search(`//*[@data-content-type="LABEL"]`)
ell, err := article.Search(`.//*[@data-content-type="LABEL"]`)
p(err)
for _, el := range ell {
s := strings.TrimSpace(el.Content())
@@ -196,7 +196,7 @@ func doArticle(filename string, url string, timestamp time.Time, needUpdate bool
pars := make([]string, 0)
ell, err = article.Search(`//*[@data-content-type="TITLE"]`)
ell, err = article.Search(`.//*[@data-content-type="TITLE"]`)
p(err)
if len(ell) != 1 {
_ = w(fmt.Errorf("found %d titles: %s", len(ell), url))
@@ -209,7 +209,7 @@ func doArticle(filename string, url string, timestamp time.Time, needUpdate bool
}
hasIntro := false
ell, err = article.Search(`//*[@data-content-type="INTRO"]`)
ell, err = article.Search(`.//*[@data-content-type="INTRO"]`)
p(err)
for _, el := range ell {
s := strings.TrimSpace(el.Content())
@@ -222,17 +222,17 @@ func doArticle(filename string, url string, timestamp time.Time, needUpdate bool
_ = w(fmt.Errorf("no intro: %s", url))
}
specials, err := article.Search(`//*[@data-content-type="GROUP"]`)
specials, err := article.Search(`.//*[@data-content-type="GROUP"]`)
p(err)
for _, special := range specials {
special.Remove()
}
other, err := article.Search(`//*[@data-content-type="PODCAST"]`)
other, err := article.Search(`.//*[@data-content-type="PODCAST"]`)
p(err)
hasOther := len(other) > 0
ell, err = article.Search(`//*[@data-content-type="PARAGRAPH"]`)
ell, err = article.Search(`.//*[@data-content-type="PARAGRAPH"]`)
p(err)
if len(ell) == 0 && !hasOther && !hasIntro {
_ = w(fmt.Errorf("no paragraphs: %s", url))

14
Parool/cmd/parool/parool.go
View File

@@ -179,7 +179,7 @@ func doArticle(filename string, url string, timestamp time.Time, needUpdate bool
}
article := articles[0]
live, err := article.Search(`//*[@data-test-id="live-blog-label"]`)
live, err := article.Search(`.//*[@data-test-id="live-blog-label"]`)
p(err)
if len(live) > 0 {
fp, err := os.Create(filename + ".skip")
@@ -189,7 +189,7 @@ func doArticle(filename string, url string, timestamp time.Time, needUpdate bool
return true
}
headers, err := article.Search(`//header`)
headers, err := article.Search(`.//header`)
p(err)
if len(headers) == 0 {
_ = w(fmt.Errorf("no header: %s", url))
@@ -212,7 +212,7 @@ func doArticle(filename string, url string, timestamp time.Time, needUpdate bool
header := headers[0]
tags := make([]string, 0)
ell, err := header.Search(`//*[@data-test-id="article-label"]`)
ell, err := header.Search(`.//*[@data-test-id="article-label"]`)
p(err)
if len(ell) == 0 {
_ = w(fmt.Errorf("no labels: %s", url))
@@ -226,7 +226,7 @@ func doArticle(filename string, url string, timestamp time.Time, needUpdate bool
pars := make([]string, 0)
ell, err = header.Search(`//*[@data-test-id="article-title"]`)
ell, err = header.Search(`.//*[@data-test-id="article-title"]`)
p(err)
if len(ell) != 1 {
_ = w(fmt.Errorf("found %d titles: %s", len(ell), url))
@@ -239,7 +239,7 @@ func doArticle(filename string, url string, timestamp time.Time, needUpdate bool
}
found := false
ell, err = header.Search(`//*[@data-test-id="header-intro"]`)
ell, err = header.Search(`.//*[@data-test-id="header-intro"]`)
p(err)
for _, el := range ell {
s := strings.TrimSpace(el.Content())
@@ -252,13 +252,13 @@ func doArticle(filename string, url string, timestamp time.Time, needUpdate bool
_ = w(fmt.Errorf("no intro: %s", url))
}
specials, err := article.Search(`//section//aside | //section//figure | //section//b`)
specials, err := article.Search(`.//section//aside | .//section//figure | .//section//b`)
p(err)
for _, special := range specials {
special.Remove()
}
ell, err = article.Search(`//section//*[@data-article-element-index]`)
ell, err = article.Search(`.//section//*[@data-article-element-index]`)
p(err)
if len(ell) == 0 {
_ = w(fmt.Errorf("no elements: %s", url))