36 lines
712 B
Go
36 lines
712 B
Go
package scraper
|
|
|
|
import (
|
|
"bytes"
|
|
"encoding/json"
|
|
"log"
|
|
"regexp"
|
|
"strings"
|
|
|
|
"github.com/PuerkitoBio/goquery"
|
|
)
|
|
|
|
func text_content(doc *goquery.Selection, selector string) string {
|
|
return cleanup_string(doc.Find(selector).First().Text())
|
|
}
|
|
|
|
var whitespace *regexp.Regexp = regexp.MustCompile(`\s+`)
|
|
|
|
func cleanup_string(s string) string {
|
|
return strings.TrimSpace(whitespace.ReplaceAllString(s, " "))
|
|
}
|
|
|
|
func document_from_bytes(b []byte) *goquery.Selection {
|
|
doc, _ := goquery.NewDocumentFromReader(bytes.NewReader(b))
|
|
return doc.Find("html").First()
|
|
}
|
|
|
|
func print_json(v any) {
|
|
out, err := json.MarshalIndent(v, "", " ")
|
|
if err != nil {
|
|
log.Println(err)
|
|
} else {
|
|
log.Println(string(out))
|
|
}
|
|
}
|