first commit

This commit is contained in:
Peter Kleiweg
2026-03-02 15:34:37 +01:00
commit 36f051a8a9
50 changed files with 4241 additions and 0 deletions

View File

@@ -0,0 +1,92 @@
package main
import (
"github.com/jbowtie/gokogiri"
"github.com/pebbe/util"
"encoding/xml"
"fmt"
"os"
"regexp"
"strings"
"time"
)
type Item struct {
Title string `xml:"title"`
Text string `xml:"description"`
}
var (
x = util.CheckErr
reYearWeek = regexp.MustCompile(`^2[0-9][0-9][0-9]-[0-5][0-9]$`)
)
func main() {
var ds string
switch len(os.Args) {
case 1:
year, week := time.Now().AddDate(0, 0, -7).ISOWeek()
ds = fmt.Sprintf("%d-%02d", year, week)
case 2:
if !reYearWeek.MatchString(os.Args[1]) {
x(fmt.Errorf("arg must be yyyy-ww"))
}
ds = os.Args[1]
default:
x(fmt.Errorf("too many arguments"))
}
dp := ds[:4] + "/" + ds[5:]
x(os.Chdir("/net/corpora/nlnieuws/AT5/" + dp))
x(os.MkdirAll("out", 0777))
files, err := os.ReadDir(".")
x(err)
for _, file := range files {
filename := file.Name()
if !strings.HasSuffix(filename, ".xml") {
continue
}
b, err := os.ReadFile(filename)
x(err)
fp, err := os.Create("out/" + filename[:len(filename)-4] + ".txt")
x(err)
var item Item
x(xml.Unmarshal(b, &item), filename)
_, err = fp.WriteString(addEnd(item.Title))
x(err)
doc, err := gokogiri.ParseHtml([]byte(`<html><body>` + item.Text + `</body></html>`))
x(err)
root := doc.Root()
pp, err := root.Search(`//body/p | //body/h2`)
x(err)
for _, p := range pp {
_, err = fp.WriteString(addEnd(p.Content()))
x(err)
}
x(err)
x(fp.Close())
}
}
func addEnd(s string) string {
s = strings.TrimSpace(s)
n := len(s)
if n == 0 {
return ""
}
if n > 0 {
if strings.ContainsAny(s[n-1:], ".!?") {
return s + "\n"
}
}
if n > 1 {
s2 := s[n-2:]
if s2 == `."` || s2 == `!"` || s2 == `?"` || s2 == `.'` || s2 == `!'` || s2 == `?'` {
return s + "\n"
}
}
return s + ".\n"
}