codeberg.org/pebbe/errors
This commit is contained in:
124
NU/cmd/nu/nu.go
124
NU/cmd/nu/nu.go
@@ -1,7 +1,7 @@
|
||||
package main
|
||||
|
||||
import (
|
||||
"github.com/pebbe/util"
|
||||
e "codeberg.org/pebbe/errors"
|
||||
|
||||
"bytes"
|
||||
"encoding/json"
|
||||
@@ -39,8 +39,8 @@ type GItem struct {
|
||||
}
|
||||
|
||||
var (
|
||||
x = util.CheckErr
|
||||
w = util.WarnErr
|
||||
p = e.PanicErr
|
||||
w = e.WarnErr
|
||||
agent = "AhrefsBot/7.0"
|
||||
)
|
||||
|
||||
@@ -57,21 +57,31 @@ 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)
|
||||
}
|
||||
}()
|
||||
|
||||
resp, err := http.Get("https://www.nu.nl/rss")
|
||||
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 {
|
||||
@@ -79,7 +89,7 @@ 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/NU/%d/%02d", year, week)
|
||||
if exists(dirname + "/lock") {
|
||||
@@ -90,22 +100,24 @@ 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, 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, t, needUpdate)
|
||||
}()
|
||||
}
|
||||
}
|
||||
|
||||
@@ -126,15 +138,15 @@ func doArticle(filename string, url string, timestamp time.Time, needUpdate bool
|
||||
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())
|
||||
|
||||
s := string(body)
|
||||
ok := true
|
||||
@@ -154,38 +166,34 @@ func doArticle(filename string, url string, timestamp time.Time, needUpdate bool
|
||||
_ = w(fmt.Errorf("script jsonld not found: %s", url))
|
||||
|
||||
fp, err := os.Create(filename + ".err")
|
||||
x(err)
|
||||
_, err = fmt.Fprintf(fp, "script jsonld not found: %s\n", url)
|
||||
x(err)
|
||||
x(fp.Close())
|
||||
x(os.Chtimes(filename+".err", timestamp, timestamp))
|
||||
p(err)
|
||||
p(fmt.Fprintf(fp, "script jsonld not found: %s\n", url))
|
||||
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
|
||||
}
|
||||
fp, err := os.Create(filename + ".json")
|
||||
x(err)
|
||||
_, err = fp.WriteString(s)
|
||||
x(err)
|
||||
x(fp.Close())
|
||||
x(os.Chtimes(filename+".json", timestamp, timestamp))
|
||||
p(err)
|
||||
p(fp.WriteString(s))
|
||||
p(fp.Close())
|
||||
p(os.Chtimes(filename+".json", timestamp, timestamp))
|
||||
|
||||
var doc Doc
|
||||
if err = json.Unmarshal([]byte(s), &doc); err != nil {
|
||||
_ = w(err, url)
|
||||
|
||||
fp, err2 := os.Create(filename + ".err")
|
||||
x(err2)
|
||||
_, err2 = fmt.Fprintf(fp, "%s: %v\n%s\n", url, err, s)
|
||||
x(err2)
|
||||
x(fp.Close())
|
||||
x(os.Chtimes(filename+".err", timestamp, timestamp))
|
||||
p(err2)
|
||||
p(fmt.Fprintf(fp, "%s: %v\n%s\n", url, err, s))
|
||||
p(fp.Close())
|
||||
p(os.Chtimes(filename+".err", timestamp, timestamp))
|
||||
|
||||
return false
|
||||
}
|
||||
@@ -193,8 +201,7 @@ func doArticle(filename string, url string, timestamp time.Time, needUpdate bool
|
||||
cats := make([]string, 0)
|
||||
var buffer bytes.Buffer
|
||||
for _, i := range doc.Graph {
|
||||
_, err = buffer.WriteString(html.UnescapeString(i.ArticleBody))
|
||||
x(err)
|
||||
p(buffer.WriteString(html.UnescapeString(i.ArticleBody)))
|
||||
cats = append(cats, i.ArticleSection...)
|
||||
}
|
||||
text := buffer.String()
|
||||
@@ -224,20 +231,17 @@ func doArticle(filename string, url string, timestamp time.Time, needUpdate bool
|
||||
text = strings.Join(lines, "\n") + "\n"
|
||||
|
||||
fp, err := os.Create(filename + ".txt")
|
||||
x(err)
|
||||
p(err)
|
||||
if len(cats) == 0 {
|
||||
_, err := fmt.Fprintln(fp, "##META text cat =")
|
||||
x(err)
|
||||
p(fmt.Fprintln(fp, "##META text cat ="))
|
||||
} else {
|
||||
for _, cat := range cats {
|
||||
_, err := fmt.Fprintf(fp, "##META text cat = %s\n", fixSpace(cat))
|
||||
x(err)
|
||||
p(fmt.Fprintf(fp, "##META text cat = %s\n", fixSpace(cat)))
|
||||
}
|
||||
}
|
||||
_, err = fp.WriteString(text)
|
||||
x(err)
|
||||
x(fp.Close())
|
||||
x(os.Chtimes(filename+".txt", timestamp, timestamp))
|
||||
p(fp.WriteString(text))
|
||||
p(fp.Close())
|
||||
p(os.Chtimes(filename+".txt", timestamp, timestamp))
|
||||
}
|
||||
|
||||
return true
|
||||
|
||||
Reference in New Issue
Block a user