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=\"VRT\"/>\n")
|
||||
x(err)
|
||||
x(fp.WriteString(s[:i]))
|
||||
x(fp.WriteString("<metadata>\n <meta type=\"text\" name=\"source\" value=\"VRT\"/>\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"
|
||||
@@ -43,8 +43,8 @@ type LinkT struct {
|
||||
}
|
||||
|
||||
var (
|
||||
x = util.CheckErr
|
||||
w = util.WarnErr
|
||||
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"
|
||||
)
|
||||
@@ -62,21 +62,31 @@ 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.vrt.be/vrtnws/nl.rss.headlines.xml")
|
||||
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 {
|
||||
@@ -84,7 +94,7 @@ func main() {
|
||||
if err != nil {
|
||||
t, err = time.Parse(time.RFC1123, item.Published)
|
||||
}
|
||||
x(err)
|
||||
p(err)
|
||||
t2, err := time.Parse(time.RFC3339Nano, item.Updated)
|
||||
if err != nil {
|
||||
t2, _ = time.Parse(time.RFC1123, item.Updated)
|
||||
@@ -102,29 +112,31 @@ 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))
|
||||
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))
|
||||
|
||||
var link string
|
||||
for _, l := range item.Link {
|
||||
if l.Type == "text/html" {
|
||||
link = l.Href
|
||||
var link string
|
||||
for _, l := range item.Link {
|
||||
if l.Type == "text/html" {
|
||||
link = l.Href
|
||||
}
|
||||
}
|
||||
}
|
||||
if !doArticle(filename, link, item.Title.Text, item.Nstag, item.Nslabeltag, t, needUpdate) {
|
||||
x(os.Remove(filename + ".xml"))
|
||||
}
|
||||
ok = doArticle(filename, link, item.Title.Text, item.Nstag, item.Nslabeltag, t, needUpdate)
|
||||
}()
|
||||
}
|
||||
}
|
||||
|
||||
@@ -142,15 +154,15 @@ func doArticle(filename string, url string, title string, tags []string, labels
|
||||
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)
|
||||
@@ -171,80 +183,71 @@ func doArticle(filename string, url string, title string, tags []string, labels
|
||||
_ = 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 buf bytes.Buffer
|
||||
|
||||
doc, err := gokogiri.ParseHtml(body)
|
||||
x(err)
|
||||
p(err)
|
||||
|
||||
root := doc.Root()
|
||||
|
||||
lnn, err := root.Search(`//head/link[@rel="canonical"]/@href`)
|
||||
x(err)
|
||||
p(err)
|
||||
for _, ln := range lnn {
|
||||
if strings.Contains(ln.String(), "/liveblog/") {
|
||||
fp, err := os.Create(filename + ".skip")
|
||||
x(err)
|
||||
_, err = fp.WriteString("liveblog\n")
|
||||
x(err)
|
||||
x(fp.Close())
|
||||
x(os.Chtimes(filename+".skip", timestamp, timestamp))
|
||||
p(err)
|
||||
p(fp.WriteString("liveblog\n"))
|
||||
p(fp.Close())
|
||||
p(os.Chtimes(filename+".skip", timestamp, timestamp))
|
||||
return true
|
||||
}
|
||||
}
|
||||
|
||||
if len(tags) == 0 {
|
||||
_, err = fmt.Fprintln(&buf, "##META text cat =")
|
||||
x(err)
|
||||
p(fmt.Fprintln(&buf, "##META text cat ="))
|
||||
} else {
|
||||
for _, tag := range tags {
|
||||
_, err = fmt.Fprintf(&buf, "##META text cat = %s\n", fixSpace(tag))
|
||||
x(err)
|
||||
p(fmt.Fprintf(&buf, "##META text cat = %s\n", fixSpace(tag)))
|
||||
}
|
||||
}
|
||||
if len(labels) == 0 {
|
||||
_, err = fmt.Fprintln(&buf, "##META text label =")
|
||||
x(err)
|
||||
p(fmt.Fprintln(&buf, "##META text label ="))
|
||||
} else {
|
||||
for _, label := range labels {
|
||||
_, err = fmt.Fprintf(&buf, "##META text label = %s\n", fixSpace(label))
|
||||
x(err)
|
||||
p(fmt.Fprintf(&buf, "##META text label = %s\n", fixSpace(label)))
|
||||
}
|
||||
}
|
||||
|
||||
_, err = buf.WriteString(addEnd(fixSpace(title)))
|
||||
x(err)
|
||||
p(err)
|
||||
|
||||
fouten := make([]string, 0)
|
||||
|
||||
found := false
|
||||
pp, err := root.Search(`//div[@data-sentry-component="ArticleHeading"]//*[contains(@class,"prose-article-body-r")]`)
|
||||
x(err)
|
||||
for _, p := range pp {
|
||||
_, err = fmt.Fprint(&buf, addEnd(fixSpace(p.Content())))
|
||||
x(err)
|
||||
p(err)
|
||||
for _, p1 := range pp {
|
||||
p(fmt.Fprint(&buf, addEnd(fixSpace(p1.Content()))))
|
||||
found = true
|
||||
}
|
||||
if !found {
|
||||
@@ -257,10 +260,9 @@ func doArticle(filename string, url string, title string, tags []string, labels
|
||||
`//div[@data-sentry-component="ArticleText"]//*[contains(@class,"prose-article-body-r")]` +
|
||||
` | ` +
|
||||
`//div[@data-sentry-component="ArticleTitle"]//h2`)
|
||||
x(err)
|
||||
for _, p := range pp {
|
||||
_, err = fmt.Fprint(&buf, addEnd(fixSpace(p.Content())))
|
||||
x(err)
|
||||
p(err)
|
||||
for _, p1 := range pp {
|
||||
p(fmt.Fprint(&buf, addEnd(fixSpace(p1.Content()))))
|
||||
found = true
|
||||
}
|
||||
if !found {
|
||||
@@ -270,31 +272,27 @@ func doArticle(filename string, url string, title string, tags []string, labels
|
||||
|
||||
if len(fouten) > 0 {
|
||||
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(err)
|
||||
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
|
||||
}
|
||||
|
||||
fp, err := os.Create(filename + ".txt")
|
||||
x(err)
|
||||
_, err = fp.Write(buf.Bytes())
|
||||
x(err)
|
||||
x(fp.Close())
|
||||
x(os.Chtimes(filename+".txt", timestamp, timestamp))
|
||||
p(err)
|
||||
p(fp.Write(buf.Bytes()))
|
||||
p(fp.Close())
|
||||
p(os.Chtimes(filename+".txt", timestamp, timestamp))
|
||||
|
||||
return true
|
||||
}
|
||||
|
||||
@@ -51,8 +51,7 @@ cd out
|
||||
mkdir xml
|
||||
Alpino -flag treebank xml debug=1 end_hook=xml user_max=900000 -parse < $corpus.lines 2> $corpus.log
|
||||
|
||||
../../../metadata 2> err
|
||||
rm err
|
||||
../../../metadata
|
||||
|
||||
cd xml
|
||||
alto -o $corpus.data.dz *.xml 2> /dev/null
|
||||
|
||||
Reference in New Issue
Block a user