Nieuw formaat van Sikkom

This commit is contained in:
Peter Kleiweg
2026-07-23 18:47:57 +02:00
parent 2466dbb43c
commit c856a372a1
4 changed files with 127 additions and 41 deletions

View File

@@ -3,14 +3,18 @@ package main
import (
e "codeberg.org/pebbe/errors"
"github.com/jbowtie/gokogiri"
"golang.org/x/net/publicsuffix"
u "git.web.rug.nl/p209327/nlnieuws/internal/util"
"bytes"
"encoding/json"
"encoding/xml"
"errors"
"fmt"
"html"
"io"
"net/http"
"net/http/cookiejar"
"net/url"
"os"
"strings"
@@ -28,13 +32,23 @@ type ItemT struct {
UnixTime int64 `xml:"unixTime"`
Guid string `xml:"guid"`
Link string `xml:"link"`
Premium bool `xml:"premium"`
Data []byte `xml:",innerxml"`
}
type LdJson struct {
Type string `json:"@type"`
IsFree string `json:"isAccessibleForFree"`
H1 string `json:"headline"`
H2 string `json:"description"`
}
var (
p = e.PanicErr
w = e.WarnErr
agent = "AhrefsBot/7.0"
// agent = "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/145.0.0.0 Safari/537.36"
jar *cookiejar.Jar
)
func exists(filename string) bool {
@@ -64,26 +78,37 @@ func main() {
}
}()
var err error
jar, err = cookiejar.New(&cookiejar.Options{PublicSuffixList: publicsuffix.List})
p(err)
myLock := "/net/corpora/nlnieuws/Sikkom/lock"
u.MkLock(myLock)
defer func() {
_ = os.Remove(myLock)
}()
resp, err := http.Get("https://www.sikkom.nl/api/feed/rss")
resp, err := http.Get("https://www.sikkom.nl/rss")
p(err)
body, err := io.ReadAll(resp.Body)
p(err)
p(resp.Body.Close())
b2 := bytes.ReplaceAll(
bytes.ReplaceAll(body, []byte("<atom:"), []byte("<atom_")),
[]byte("</atom:"), []byte("</atom_"))
var rss Rss
p(xml.Unmarshal(body, &rss))
p(xml.Unmarshal(b2, &rss))
if len(rss.Items) == 0 {
p(fmt.Errorf("len(rss.Items) == 0"))
}
for _, item := range rss.Items {
if item.Premium {
continue
}
t, err := time.Parse(time.RFC1123Z, item.PubDate)
if err != nil {
t, err = time.Parse(time.RFC1123, item.PubDate)
@@ -145,7 +170,7 @@ func doArticle(filename string, url string, title string, timestamp time.Time, n
p(err)
req.Header.Set("User-Agent", agent)
client := &http.Client{}
client := &http.Client{Jar: jar}
resp, err := client.Do(req)
p(err)
body, err := io.ReadAll(resp.Body)
@@ -154,22 +179,31 @@ func doArticle(filename string, url string, title string, timestamp time.Time, n
body = u.HtmlFix(body)
s := string(body)
doc, err := gokogiri.ParseHtml(body)
p(err)
ok := true
i1 := strings.Index(s, `"application/ld+json"`)
if i1 < 0 {
ok = false
} else {
i1 += strings.Index(s[i1:], `>`) + 1
i2 := i1 + strings.Index(s[i1:], `</script>`)
if i2 < i1 {
ok = false
} else {
s = html.UnescapeString(s[i1:i2])
root := doc.Root()
ldjsons, err := root.Search(`//script[@type="application/ld+json"]`)
p(err)
found := false
var jsmain LdJson
for _, ldjson := range ldjsons {
s := ldjson.Content()
var js LdJson
p(json.Unmarshal([]byte(s), &js))
if js.Type == "NewsArticle" {
found = true
fp, err := os.Create(filename + ".json")
p(err)
p(fp.WriteString(s))
p(fp.Close())
p(os.Chtimes(filename+".json", timestamp, timestamp))
jsmain = js
break
}
}
if !ok {
if !found {
_ = w(fmt.Errorf("script jsonld not found: %s", url))
fp, err := os.Create(filename + ".err")
@@ -183,30 +217,24 @@ func doArticle(filename string, url string, title string, timestamp time.Time, n
p(fp.Write(body))
p(fp.Close())
p(os.Chtimes(filename+".html", timestamp, timestamp))
return false
}
fp, err := os.Create(filename + ".json")
locations, err := root.Search(`//span[starts-with(@class, "location")]`)
p(err)
p(fp.WriteString(s))
p(fp.Close())
p(os.Chtimes(filename+".json", timestamp, timestamp))
for _, location := range locations {
location.Remove()
}
doc, err := gokogiri.ParseHtml(body)
articles, err := root.Search(`//article[starts-with(@class, "article")]`)
p(err)
root := doc.Root()
pp, err := root.Search(`//div[contains(@class,"article-page__body")]//p`)
p(err)
if len(pp) == 0 {
_ = w(fmt.Errorf("empty: %s", url))
if len(articles) != 1 {
msg := fmt.Sprintf("len(articles) == %d in %s", len(articles), url)
_ = w(errors.New(msg))
fp, err := os.Create(filename + ".err")
p(err)
p(fmt.Fprintf(fp, "empty: %s\n", url))
p(fmt.Fprintln(fp, msg))
p(fp.Close())
p(os.Chtimes(filename+".err", timestamp, timestamp))
@@ -220,15 +248,57 @@ func doArticle(filename string, url string, title string, timestamp time.Time, n
return false
}
fp, err = os.Create(filename + ".txt")
p(err)
art := articles[0]
p(fp.WriteString(u.AddEnd(u.FixSpace(title))))
for _, p1 := range pp {
p(fp.WriteString(u.AddEnd(u.FixSpace(p1.Content()))))
text := make([]string, 0)
if jsmain.H1 != "" {
text = append(text, jsmain.H1)
} else {
text = append(text, title)
}
if jsmain.H2 != "" {
text = append(text, jsmain.H2)
}
abb, err := art.Search(`.//div[starts-with(@class, "article-body")]`)
p(err)
found = false
for _, ab := range abb {
ell, err := ab.Search(`.//p | .//h4`)
p(err)
for _, el := range ell {
if el.Name() == "p" {
found = true
}
text = append(text, el.Content())
}
}
if !found {
_ = w(fmt.Errorf("empty body: %s", url))
fp, err := os.Create(filename + ".err")
p(err)
p(fmt.Fprintf(fp, "empty body: %s\n", url))
p(fp.Close())
p(os.Chtimes(filename+".err", timestamp, timestamp))
fp, err = os.Create(filename + ".html")
p(err)
p(fp.Write(body))
p(fp.Close())
p(os.Chtimes(filename+".html", timestamp, timestamp))
return false
}
fp, err := os.Create(filename + ".txt")
p(err)
for _, t := range text {
p(fp.WriteString(u.AddEnd(u.FixSpace(t))))
}
p(fp.Close())
return true
}

8
go.mod
View File

@@ -5,9 +5,15 @@ go 1.26.1
require (
codeberg.org/pebbe/errors v0.4.0
github.com/jbowtie/gokogiri v0.0.0-20250107075044-de0f9d4877a5
github.com/kr/pretty v0.3.1
github.com/pebbe/compactcorpus v1.0.3
github.com/pebbe/textcat/v2 v2.3.0
github.com/rug-compling/alpinods v1.18.1
golang.org/x/net v0.57.0
)
require github.com/pebbe/util v0.9.0 // indirect
require (
github.com/kr/text v0.2.0 // indirect
github.com/pebbe/util v0.9.0 // indirect
github.com/rogpeppe/go-internal v1.9.0 // indirect
)

10
go.sum
View File

@@ -1,12 +1,22 @@
codeberg.org/pebbe/errors v0.4.0 h1:G05wsXpC/LRPaL02QYDwtz0sWFWQcIWK1s+MC79LBzU=
codeberg.org/pebbe/errors v0.4.0/go.mod h1:O7PPxUJM1bWRHq11CRK3wqVaH/3NnRaSVZvh3UhzDCY=
github.com/creack/pty v1.1.9/go.mod h1:oKZEueFk5CKHvIhNR5MUki03XCEU+Q6VDXinZuGJ33E=
github.com/jbowtie/gokogiri v0.0.0-20250107075044-de0f9d4877a5 h1:tQbR4RKFBFi0+Ll69dXejKKUbQVNaOAT2fjlDvSAfx4=
github.com/jbowtie/gokogiri v0.0.0-20250107075044-de0f9d4877a5/go.mod h1:kQE2lxPgVKe0JsBZMFFfMm5zBDCuRhaHFKOBzZeCLiw=
github.com/kr/pretty v0.3.1 h1:flRD4NNwYAUpkphVc1HcthR4KEIFJ65n8Mw5qdRn3LE=
github.com/kr/pretty v0.3.1/go.mod h1:hoEshYVHaxMs3cyo3Yncou5ZscifuDolrwPKZanG3xk=
github.com/kr/text v0.2.0 h1:5Nx0Ya0ZqY2ygV366QzturHI13Jq95ApcVaJBhpS+AY=
github.com/kr/text v0.2.0/go.mod h1:eLer722TekiGuMkidMxC/pM04lWEeraHUUmBw8l2grE=
github.com/pebbe/compactcorpus v1.0.3 h1:6qlfXKHTKg7oWKLPCgEgv1scplfvphg/9l9XiRT2HzQ=
github.com/pebbe/compactcorpus v1.0.3/go.mod h1:SSpTeCZataCjjs82RJb8SOGdjkB3PlR7Z19EY4rInoQ=
github.com/pebbe/textcat/v2 v2.3.0 h1:RB2egIQgI2a2Ls+I9No6KFQKCZBIFt8Cc/SWCnVtC7Y=
github.com/pebbe/textcat/v2 v2.3.0/go.mod h1:WLXWuL+fOlQJqn6LmubjD+e78hCC6Y/rAWInh0wq/kg=
github.com/pebbe/util v0.9.0 h1:PMZd+CpWb8GbWEmFGlL3qd6XPuywl6xFIbrXWi870OA=
github.com/pebbe/util v0.9.0/go.mod h1:ynWl/SFX4+Seb9fpjVlYevr1f4TP7FrCmyZHiBCg69Q=
github.com/pkg/diff v0.0.0-20210226163009-20ebb0f2a09e/go.mod h1:pJLUxLENpZxwdsKMEsNbx1VGcRFpLqf3715MtcvvzbA=
github.com/rogpeppe/go-internal v1.9.0 h1:73kH8U+JUqXU8lRuOHeVHaa/SZPifC7BkcraZVejAe8=
github.com/rogpeppe/go-internal v1.9.0/go.mod h1:WtVeX8xhTBvf0smdhujwtBcq4Qrzq/fJaraNFVN+nFs=
github.com/rug-compling/alpinods v1.18.1 h1:BvPcCnNEQ1QoVSc0RmwJd3kZmvo4iqZ52/vFzVvFS7w=
github.com/rug-compling/alpinods v1.18.1/go.mod h1:R3BBX8RIw9InVqHZ+1W+MsX8WX8uBkoVNNGE38mqF1Q=
golang.org/x/net v0.57.0 h1:K5+3DljvIuDG9/Jv9rvyMywYNFCQ9RSUY6OOTTkT+tE=
golang.org/x/net v0.57.0/go.mod h1:KpXc8iv+r3XplLAG/f7Jsf9RPszJzdR0f58q9vGOuEU=

View File

@@ -1,7 +1,7 @@
<!DOCTYPE html>
<html>
<head>
<title id="title">Woord van de maand</title>
<title id="title">Stijger van de week</title>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1" />
<link rel="icon" href="favicon.ico" type="image/ico" />
@@ -12,7 +12,7 @@
</head>
<body>
<div class="title">
<h1>Woord van de maand</h1>
<h1>Stijger van de week</h1>
</div>
<div class="form">
<form name="choice">