goParser

parser.go

package html

import (
	"golang.org/x/net/html"
	"net/http"
)

type Filter func(tagName string) bool
type Extractor func(node *html.Node) Data
type Data any

type Parser interface {
	Import(url string) ([]Data, error)
}

type parser struct {
	filter    Filter
	extractor Extractor
}

func (p parser) Import(url string) ([]Data, error) {
	ret, err := http.Get(url)
	if err != nil {
		return nil, err
	}
	defer ret.Body.Close()
	data, err := html.Parse(ret.Body)
	if err != nil {
		return nil, err
	}

	result := make([]Data, 0)
	for n := range data.Descendants() {
		if n.Type == html.ElementNode && p.filter(n.Data) {
			d := p.extractor(n)
			if d != nil {
				result = append(result, d)
			}
		}
	}
	return result, nil
}

func NewParser(filter Filter, extractor Extractor) Parser {
	return &parser{
		filter:    filter,
		extractor: extractor,
	}
}

parser_test.go

package html

import (
	"fmt"
	"golang.org/x/net/html"
	"testing"
)

func TestParser(t *testing.T) {
	p := NewParser(func(tagName string) bool {
		return tagName == "img"
	}, func(node *html.Node) Data {
		for i := range node.Attr {
			attr := node.Attr[i]
			if attr.Key == "src" {
				return attr.Val
			}
		}
		return nil
	})
	ret, _ := p.Import("https://skillfactory.ru")
	for i := range ret {
		href := ret[i].(string)
		fmt.Println(href)
	}
}