codeberg.org/pebbe/errors
This commit is contained in:
108
GG/cmd/gg/gg.go
108
GG/cmd/gg/gg.go
@@ -1,8 +1,8 @@
|
||||
package main
|
||||
|
||||
import (
|
||||
e "codeberg.org/pebbe/errors"
|
||||
"github.com/jbowtie/gokogiri"
|
||||
"github.com/pebbe/util"
|
||||
|
||||
"encoding/xml"
|
||||
"fmt"
|
||||
@@ -29,8 +29,8 @@ type ItemT 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"
|
||||
)
|
||||
@@ -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://gemeente.groningen.nl/feed/rss/nieuws", 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)
|
||||
year, week := t.ISOWeek()
|
||||
dirname := fmt.Sprintf("/net/corpora/nlnieuws/GG/%d/%02d", year, week)
|
||||
if exists(dirname + "/lock") {
|
||||
@@ -86,26 +96,28 @@ 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)
|
||||
}()
|
||||
}
|
||||
}
|
||||
|
||||
func doArticle(filename string, url string, title string, timestamp time.Time, needUpdate bool) bool {
|
||||
func doArticle(filename string, url string, title string, timestamp time.Time, needUpdate bool) (ok bool) {
|
||||
if needUpdate {
|
||||
_ = os.Remove(filename + ".err")
|
||||
_ = os.Remove(filename + ".html")
|
||||
@@ -119,18 +131,18 @@ 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)
|
||||
|
||||
root := doc.Root()
|
||||
|
||||
@@ -138,41 +150,37 @@ func doArticle(filename string, url string, title string, timestamp time.Time, n
|
||||
`//div[contains(@class,"component-richtext")]/p` +
|
||||
` | ` +
|
||||
`//div[contains(@class,"component-richtext")]/h2`)
|
||||
x(err)
|
||||
p(err)
|
||||
if len(ell) == 0 {
|
||||
_ = w(fmt.Errorf("empty: %s", url))
|
||||
|
||||
fp, err := os.Create(filename + ".err")
|
||||
x(err)
|
||||
_, err = fmt.Fprintf(fp, "empty: %s\n", url)
|
||||
x(err)
|
||||
x(fp.Close())
|
||||
x(os.Chtimes(filename+".err", timestamp, timestamp))
|
||||
p(err)
|
||||
p(fmt.Fprintf(fp, "empty: %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 + ".txt")
|
||||
x(err)
|
||||
p(err)
|
||||
|
||||
_, err = fp.WriteString(addEnd(fixSpace(title)))
|
||||
x(err)
|
||||
p(fp.WriteString(addEnd(fixSpace(title))))
|
||||
|
||||
for _, el := range ell {
|
||||
_, err = fp.WriteString(addEnd(fixSpace(el.Content())))
|
||||
x(err)
|
||||
p(fp.WriteString(addEnd(fixSpace(el.Content()))))
|
||||
}
|
||||
|
||||
x(fp.Close())
|
||||
p(fp.Close())
|
||||
|
||||
x(os.Chtimes(filename+".txt", timestamp, timestamp))
|
||||
p(os.Chtimes(filename+".txt", timestamp, timestamp))
|
||||
|
||||
return true
|
||||
}
|
||||
|
||||
@@ -1,7 +1,7 @@
|
||||
package main
|
||||
|
||||
import (
|
||||
"github.com/pebbe/util"
|
||||
e "codeberg.org/pebbe/errors"
|
||||
|
||||
"encoding/xml"
|
||||
"fmt"
|
||||
@@ -16,7 +16,7 @@ type Item struct {
|
||||
}
|
||||
|
||||
var (
|
||||
x = util.CheckErr
|
||||
x = e.ExitErr
|
||||
data = make(map[string][]string)
|
||||
location *time.Location
|
||||
)
|
||||
@@ -51,18 +51,13 @@ 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=\"Gemeente Groningen\"/>\n")
|
||||
x(err)
|
||||
x(fp.WriteString(s[:i]))
|
||||
x(fp.WriteString("<metadata>\n <meta type=\"text\" name=\"source\" value=\"Gemeente Groningen\"/>\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))
|
||||
}
|
||||
|
||||
@@ -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