codeberg.org/pebbe/errors
This commit is contained in:
@@ -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
|
||||
@@ -56,18 +56,13 @@ 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=\"NieuwsNL\"/>\n")
|
||||
x(err)
|
||||
x(fp.WriteString(s[:i]))
|
||||
x(fp.WriteString("<metadata>\n <meta type=\"text\" name=\"source\" value=\"NieuwsNL\"/>\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))
|
||||
}
|
||||
|
||||
@@ -1,8 +1,8 @@
|
||||
package main
|
||||
|
||||
import (
|
||||
e "codeberg.org/pebbe/errors"
|
||||
"github.com/jbowtie/gokogiri"
|
||||
"github.com/pebbe/util"
|
||||
|
||||
"bytes"
|
||||
"encoding/xml"
|
||||
@@ -30,8 +30,8 @@ type ItemT struct {
|
||||
}
|
||||
|
||||
var (
|
||||
x = util.CheckErr
|
||||
w = util.WarnErr
|
||||
p = e.PanicErr
|
||||
w = e.WarnErr
|
||||
agent = "AhrefsBot/7.0"
|
||||
)
|
||||
|
||||
@@ -48,26 +48,36 @@ func fileDate(filename string) string {
|
||||
s := string(b)
|
||||
i1 := strings.Index(s, "<unixTime>") + 10
|
||||
i2 := strings.Index(s, "</unixTime>")
|
||||
if i2 < i1 {
|
||||
return ""
|
||||
}
|
||||
return s[i1:i2]
|
||||
}
|
||||
|
||||
func main() {
|
||||
defer func() {
|
||||
if e.Panicked() {
|
||||
_ = recover()
|
||||
os.Exit(1)
|
||||
}
|
||||
}()
|
||||
|
||||
req, err := http.NewRequest("GET", "https://nieuws.nl/sitemap/news.xml", 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 {
|
||||
@@ -75,7 +85,7 @@ func main() {
|
||||
if err != nil {
|
||||
t, err = time.Parse(time.RFC1123, item.PubDate)
|
||||
}
|
||||
x(err)
|
||||
p(err)
|
||||
dirname := fmt.Sprintf("/net/corpora/nlnieuws/NieuwsNL/%d/%02d/%02d", t.Year(), int(t.Month()), t.Day())
|
||||
if exists(dirname + "/lock") {
|
||||
continue
|
||||
@@ -85,22 +95,24 @@ func main() {
|
||||
ts := fmt.Sprintf("%d", t.Unix())
|
||||
needUpdate := fileDate(filename+".xml") != ts
|
||||
|
||||
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))
|
||||
if !doArticle(filename, item.Link, item.Title, t, needUpdate) {
|
||||
x(os.Remove(filename + ".xml"))
|
||||
}
|
||||
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 = doArticle(filename, item.Link, item.Title, t, needUpdate)
|
||||
}()
|
||||
}
|
||||
}
|
||||
|
||||
@@ -118,18 +130,18 @@ func doArticle(filename string, url string, title string, timestamp time.Time, n
|
||||
time.Sleep(2 * time.Second)
|
||||
|
||||
req, err := http.NewRequest("GET", url, 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())
|
||||
|
||||
doc, err := gokogiri.ParseHtml(body)
|
||||
x(err)
|
||||
p(err)
|
||||
|
||||
var buf bytes.Buffer
|
||||
fouten := make([]string, 0)
|
||||
@@ -138,10 +150,9 @@ func doArticle(filename string, url string, title string, timestamp time.Time, n
|
||||
|
||||
var cat string
|
||||
aa, err := root.Search(`//a[contains(@class, "articleHeader__info__category")]`)
|
||||
x(err)
|
||||
p(err)
|
||||
if len(aa) == 0 {
|
||||
_, err = fmt.Fprintln(&buf, "##META text cat =")
|
||||
x(err)
|
||||
p(fmt.Fprintln(&buf, "##META text cat ="))
|
||||
_ = w(fmt.Errorf("no cat: %s", url))
|
||||
// geen fout, maar waarschuwing als er meer fouten zijn
|
||||
fouten = append(fouten, fmt.Sprintf("no text: %s\n", url))
|
||||
@@ -149,52 +160,46 @@ func doArticle(filename string, url string, title string, timestamp time.Time, n
|
||||
} else {
|
||||
for _, a := range aa {
|
||||
cat = strings.ReplaceAll(a.Content(), "\n", " ")
|
||||
_, err = fmt.Fprintf(&buf, "##META text cat = %s\n", fixSpace(cat))
|
||||
x(err)
|
||||
p(fmt.Fprintf(&buf, "##META text cat = %s\n", fixSpace(cat)))
|
||||
}
|
||||
}
|
||||
|
||||
_, err = buf.WriteString(addEnd(fixSpace(title)))
|
||||
x(err)
|
||||
p(buf.WriteString(addEnd(fixSpace(title))))
|
||||
|
||||
// oud: //div[@id="article-blocks"]//p
|
||||
pp, err := root.Search(`//div[@id="article-blocks"]//div[contains(@class, "paragraph-content")]`)
|
||||
x(err)
|
||||
p(err)
|
||||
if len(pp) == 0 {
|
||||
_ = w(fmt.Errorf("empty: %s", url))
|
||||
// dit is echt fout
|
||||
fouten = append(fouten, fmt.Sprintf("empty: %s\n", url))
|
||||
|
||||
fp, err := os.Create(filename + ".err")
|
||||
x(err)
|
||||
p(err)
|
||||
for _, fout := range fouten {
|
||||
_, err = fp.WriteString(fout)
|
||||
x(err)
|
||||
p(fp.WriteString(fout))
|
||||
}
|
||||
x(fp.Close())
|
||||
x(os.Chtimes(filename+".err", timestamp, timestamp))
|
||||
p(fp.Close())
|
||||
p(os.Chtimes(filename+".err", timestamp, timestamp))
|
||||
|
||||
fp, err = os.Create(filename + ".html")
|
||||
x(err)
|
||||
_, err = fp.Write(body)
|
||||
x(err)
|
||||
x(fp.Close())
|
||||
x(os.Chtimes(filename+".html", timestamp, timestamp))
|
||||
p(err)
|
||||
p(fp.Write(body))
|
||||
p(fp.Close())
|
||||
p(os.Chtimes(filename+".html", timestamp, timestamp))
|
||||
|
||||
return false // echt fout
|
||||
}
|
||||
for _, p := range pp {
|
||||
_, err = buf.WriteString(addEnd(fixSpace(p.Content())))
|
||||
x(err)
|
||||
for _, p1 := range pp {
|
||||
p(buf.WriteString(addEnd(fixSpace(p1.Content()))))
|
||||
}
|
||||
|
||||
fp, err := os.Create(filename + ".txt")
|
||||
x(err)
|
||||
_, err = fp.Write(buf.Bytes())
|
||||
x(err)
|
||||
x(fp.Close())
|
||||
p(err)
|
||||
p(fp.Write(buf.Bytes()))
|
||||
p(fp.Close())
|
||||
|
||||
x(os.Chtimes(filename+".txt", timestamp, timestamp))
|
||||
p(os.Chtimes(filename+".txt", timestamp, timestamp))
|
||||
|
||||
return true
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user