Files
nlnieuws/RO/cmd/ro/ro.go
Peter Kleiweg 36f051a8a9 first commit
2026-03-02 15:34:37 +01:00

82 lines
1.7 KiB
Go

package main
import (
"github.com/pebbe/util"
"encoding/xml"
"fmt"
"io"
"net/http"
"net/url"
"os"
"strings"
"time"
)
type Rss struct {
XMLName xml.Name `xml:"rss"`
Items []ItemT `xml:"channel>item"`
}
type ItemT struct {
PubDate string `xml:"pubDate"`
UnixTime int64 `xml:"unixTime"`
Guid string `xml:"guid"`
Data []byte `xml:",innerxml"`
}
var (
x = util.CheckErr
agent = "AhrefsBot/7.0"
)
func main() {
req, err := http.NewRequest("GET", "https://reportersonline.nl/feed/", nil)
x(err)
req.Header.Set("User-Agent", agent)
client := &http.Client{}
resp, err := client.Do(req)
x(err)
body, err := io.ReadAll(resp.Body)
x(err)
x(resp.Body.Close())
var rss Rss
x(xml.Unmarshal(body, &rss))
if len(rss.Items) == 0 {
x(fmt.Errorf("len(rss.Items) == 0"))
}
for _, item := range rss.Items {
t, err := time.Parse(time.RFC1123Z, item.PubDate)
if err != nil {
t, err = time.Parse(time.RFC1123, item.PubDate)
}
x(err)
year, week := t.ISOWeek()
dirname := fmt.Sprintf("/net/corpora/nlnieuws/RO/%d/%02d", year, week)
basename := strings.TrimPrefix(item.Guid, "https://reportersonline.nl/?p=")
if i := strings.LastIndex(basename, "/"); i > 0 {
basename = basename[:i]
}
filename := dirname + "/" + url.PathEscape(basename)
x(os.MkdirAll(dirname, 0777))
fp, err := os.Create(filename + ".xml")
x(err)
_, err = fp.WriteString("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<item>\n")
x(err)
_, err = fmt.Fprintf(fp, "<unixTime>%d</unixTime>", t.Unix())
x(err)
_, err = fp.Write(item.Data)
x(err)
_, err = fp.WriteString("</item>\n")
x(err)
x(fp.Close())
x(os.Chtimes(filename+".xml", t, t))
}
}