rss/bbclearn/bbc.go

package bbclearn

import (
	"fmt"
	"github.com/PuerkitoBio/goquery"
	strings2 "github.com/fthvgb1/wp-go/helper/strings"
	"github.com/fthvgb1/wp-go/rss2"
	"net/http"
	"strings"
	"time"
)

func LearnParse(s string, recentDay int) string {
	document, err := goquery.NewDocumentFromReader(strings.NewReader(s))
	if err != nil {
		return ""
	}
	var item []rss2.Item
	item = append(item, full(document, recentDay))
	item = append(item, items(document, recentDay)...)
	if len(item) < 1 {
		return ""
	}
	rss := rss2.Rss2{
		Title:         "BBC 英语教学",
		Link:          "https://www.bbc.co.uk/learningenglish/chinese/",
		LastBuildDate: time.Now().Format(time.RFC1123Z),
		Items:         item,
	}
	return rss.GetXML()
}

func dateFilter(u string, recentDay int) (r bool) {
	uu := strings.Split(u, "-")
	if len(uu) < 2 {
		return
	}
	date, err := time.Parse("060102", uu[len(uu)-1])
	if err != nil {
		return
	}
	t := time.Now()
	if t.Year() != date.Year() || t.Month() != date.Month() {
		return
	}
	fmt.Println(time.Now().Day()-recentDay, date.Day())
	if t.Day()-recentDay > date.Day() {
		return
	}
	r = true
	return
}

func fetch(u string) (r rss2.Item) {
	res, err := http.Get(u)
	if err != nil {
		return
	}
	dom, err := goquery.NewDocumentFromReader(res.Body)
	if err != nil {
		return
	}
	s := dom.Find("#bbcle-content .widget-container-left")
	content, err := goquery.OuterHtml(s.Find(".widget-list,.widget-pagelink").Remove().End())
	if err != nil {
		return
	}
	r.Title = s.Find(`div[data-widget-index="3"] h3`).Text()

	r.PubDate = strings.TrimSpace(s.Find(".widget-bbcle-featuresubheader").Text())
	r.PubDate = strings2.Replace(r.PubDate, map[string]string{
		"\n": "",
	})
	r.PubDate = strings2.CutSpecialDuplicate(r.PubDate, " ")
	r.Guid = u
	r.Description = content
	return
}

func full(doc *goquery.Document, recentDay int) (r rss2.Item) {
	a := doc.Find("#bbcle-content .widget-container-full a")
	u, ok := a.Attr("href")
	if !ok {
		return
	}
	if !dateFilter(u, recentDay) {
		return
	}
	r = fetch(u)
	return
}

func items(doc *goquery.Document, recentDay int) (r []rss2.Item) {
	doc.Find("#bbcle-content > div > div.widget-container.widget-container-full > div.widget.widget-image.widget-image-two_column > div a").Each(func(i int, s *goquery.Selection) {
		u, ok := s.Attr("href")
		if !ok || !dateFilter(u, recentDay) {
			return
		}
		r = append(r, fetch(u))
	})
	return
}