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

codeberg.org/pebbe/errors

Browse Source
This commit is contained in:
Peter Kleiweg
2026-03-16 13:41:03 +01:00
parent 7b18c51567
commit 78dc580c8d
40 changed files with 706 additions and 692 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
Tzum/cmd/metadata/metadata.go
View File

@@ -1,7 +1,7 @@
package main
import (
"github.com/pebbe/util"
e "codeberg.org/pebbe/errors"
"bufio"
"encoding/xml"
@@ -18,7 +18,7 @@ type Item struct {
}
var (
x = util.CheckErr
x = e.ExitErr
escape = html.EscapeString
data = make(map[string][]string)
location *time.Location
@@ -66,18 +66,13 @@ func main() {
i += strings.Index(s[i:], "<")
fp, err := os.Create("xml/" + filename + ".tmp")
x(err)
_, err = fp.WriteString(s[:i])
x(err)
_, err = fp.WriteString("<metadata>\n <meta type=\"text\" name=\"source\" value=\"Tzum\"/>\n")
x(err)
x(fp.WriteString(s[:i]))
x(fp.WriteString("<metadata>\n <meta type=\"text\" name=\"source\" value=\"Tzum\"/>\n"))
for _, m := range data[base] {
_, err = fp.WriteString(" " + m + "\n")
x(err)
x(fp.WriteString(" " + m + "\n"))
}
_, err = fp.WriteString(" </metadata>\n ")
x(err)
_, err = fp.WriteString(stripMeta(s[i:]))
x(err)
x(fp.WriteString(" </metadata>\n "))
x(fp.WriteString(stripMeta(s[i:])))
x(fp.Close())
x(os.Rename("xml/"+filename+".tmp", "xml/"+filename))
}

56
Tzum/cmd/tzum/tzum.go
View File

@@ -1,7 +1,7 @@
package main
import (
"github.com/pebbe/util"
e "codeberg.org/pebbe/errors"
"encoding/xml"
"fmt"
@@ -26,7 +26,7 @@ type ItemT struct {
}
var (
x = util.CheckErr
p = e.PanicErr
agent = "AhrefsBot/7.0"
)
@@ -36,22 +36,29 @@ func exists(filename string) bool {
}
func main() {
defer func() {
if e.Panicked() {
_ = recover()
os.Exit(1)
}
}()
req, err := http.NewRequest("GET", "https://www.tzum.info/feed/", nil)
x(err)
p(err)
req.Header.Set("User-Agent", agent)
client := &http.Client{}
resp, err := client.Do(req)
x(err)
p(err)
body, err := io.ReadAll(resp.Body)
x(err)
x(resp.Body.Close())
p(err)
p(resp.Body.Close())
var rss Rss
x(xml.Unmarshal(body, &rss))
p(xml.Unmarshal(body, &rss))
if len(rss.Items) == 0 {
x(fmt.Errorf("len(rss.Items) == 0"))
p(fmt.Errorf("len(rss.Items) == 0"))
}
for _, item := range rss.Items {
@@ -59,7 +66,7 @@ func main() {
if err != nil {
t, err = time.Parse(time.RFC1123, item.PubDate)
}
x(err)
p(err)
year, week := t.ISOWeek()
dirname := fmt.Sprintf("/net/corpora/nlnieuws/Tzum/%d/%02d", year, week)
if exists(dirname + "/lock") {
@@ -71,19 +78,24 @@ func main() {
}
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))
p(os.MkdirAll(dirname, 0777))
func() {
var ok bool
defer func() {
if !ok {
_ = os.Remove(filename + ".xml")
}
}()
fp, err := os.Create(filename + ".xml")
p(err)
p(fp.WriteString("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<item>\n"))
p(fmt.Fprintf(fp, "<unixTime>%d</unixTime>", t.Unix()))
p(fp.Write(item.Data))
p(fp.WriteString("</item>\n"))
p(fp.Close())
p(os.Chtimes(filename+".xml", t, t))
ok = true
}()
}
}

14
Tzum/cmd/xml2txt/xml2txt.go
View File

@@ -1,8 +1,8 @@
package main
import (
e "codeberg.org/pebbe/errors"
"github.com/jbowtie/gokogiri"
"github.com/pebbe/util"
"encoding/xml"
"fmt"
@@ -19,7 +19,7 @@ type Item struct {
}
var (
x = util.CheckErr
x = e.ExitErr
reYearWeek = regexp.MustCompile(`^2[0-9][0-9][0-9]-[0-5][0-9]$`)
)
@@ -57,11 +57,9 @@ func main() {
var item Item
x(xml.Unmarshal(b, &item))
for _, cat := range item.Cats {
_, err = fmt.Fprintf(fp, "##META text cat = %s\n", fixSpace(cat))
x(err)
x(fmt.Fprintf(fp, "##META text cat = %s\n", fixSpace(cat)))
}
_, err = fp.WriteString(addEnd(fixSpace(item.Title)))
x(err)
x(fp.WriteString(addEnd(fixSpace(item.Title))))
doc, err := gokogiri.ParseHtml([]byte(`<html><body>` + item.Text + `</body></html>`))
x(err)
root := doc.Root()
@@ -70,11 +68,9 @@ func main() {
for _, p := range pp {
s := p.Content()
if !strings.Contains(s, "verscheen eerst op Tzum.") {
_, err = fp.WriteString(addEnd(fixSpace(p.Content())))
x(err)
x(fp.WriteString(addEnd(fixSpace(p.Content()))))
}
}
x(err)
x(fp.Close())
}
}