init

.gitignore

@@ -1,3 +1,5 @@
 .idea
 .gitignore
-fetchdapenti.iml
+rss.iml
+rss
+fetchdapenti

Dockerfile

@@ -0,0 +1,3 @@
+FROM alpine:latest
+COPY rss /opt/rss
+CMD ["/opt/rss"]

bbclearn/bbc.go

@@ -0,0 +1,99 @@
+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)...)
+	rss := rss2.Rss2{
+		Title:         "BBC 英语教学",
+		Link:          "http://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
+}

bbclearn/bbc_test.go

@@ -0,0 +1,70 @@
+package bbclearn
+
+import (
+	"github.com/PuerkitoBio/goquery"
+	"github.com/fthvgb1/wp-go/rss2"
+	"os"
+	"reflect"
+	"strings"
+	"testing"
+)
+
+func Test_full(t *testing.T) {
+	type args struct {
+		doc       *goquery.Document
+		recentDay int
+	}
+	tests := []struct {
+		name string
+		args args
+		want rss2.Item
+	}{
+		{
+			name: "t1",
+			args: args{doc: func() *goquery.Document {
+				f, _ := os.ReadFile("b.html")
+				d, _ := goquery.NewDocumentFromReader(strings.NewReader(string(f)))
+				return d
+			}(), recentDay: 1},
+			//want: "",
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := full(tt.args.doc, tt.args.recentDay); got != tt.want {
+				t.Errorf("full() = %v, want %v", got, tt.want)
+			}
+		})
+	}
+}
+
+func Test_items(t *testing.T) {
+	type args struct {
+		doc       *goquery.Document
+		recentDay int
+	}
+	tests := []struct {
+		name  string
+		args  args
+		wantR rss2.Item
+	}{
+		{
+			name: "t1",
+			args: args{
+				doc: func() *goquery.Document {
+					f, _ := os.ReadFile("b.html")
+					d, _ := goquery.NewDocumentFromReader(strings.NewReader(string(f)))
+					return d
+				}(),
+				recentDay: 1,
+			},
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if gotR := items(tt.args.doc, tt.args.recentDay); !reflect.DeepEqual(gotR, tt.wantR) {
+				t.Errorf("items() = %v, want %v", gotR, tt.wantR)
+			}
+		})
+	}
+}

go.mod

@@ -1,3 +1,11 @@
-module fetchdapenti
+module rss
 
 go 1.19
+
+require (
+	github.com/PuerkitoBio/goquery v1.8.1 // indirect
+	github.com/andybalholm/cascadia v1.3.1 // indirect
+	github.com/fthvgb1/wp-go v0.0.0-20230411054214-125764711dbd // indirect
+	golang.org/x/exp v0.0.0-20230203172020-98cc5a0785f9 // indirect
+	golang.org/x/net v0.7.0 // indirect
+)

main.go

@@ -0,0 +1,94 @@
+package main
+
+import (
+	"io"
+	"log"
+	"net/http"
+	"os"
+	"regexp"
+	"rss/bbclearn"
+	"strings"
+	"time"
+)
+
+var zhihuReg = regexp.MustCompile(`(?is:(<item>(.*?)</item>))`)
+var date = regexp.MustCompile(`<pubDate>(.*)</pubDate>`)
+
+func fetch(u string, fn ...func(s string) string) string {
+	res, err := http.Get(u)
+	if err != nil {
+		return ""
+	}
+	s, err := io.ReadAll(res.Body)
+	if err != nil {
+		return ""
+	}
+	html := string(s)
+	for _, f := range fn {
+		html = f(html)
+	}
+	return html
+}
+
+func dayLimit(today, forwardDay int, s string) string {
+	da := date.FindStringSubmatch(s)
+	if len(da) <= 1 {
+		return s
+	}
+	t, err := time.Parse(time.RFC1123Z, da[1])
+	if err != nil {
+		return s
+	}
+	if today-forwardDay > t.Day() {
+		return ""
+	}
+	return s
+}
+
+func filterItem(html string, today, recentDay int) string {
+	return zhihuReg.ReplaceAllStringFunc(html, func(s string) string {
+		return dayLimit(today, recentDay, s)
+	})
+}
+
+func penti(w http.ResponseWriter, req *http.Request) {
+	io.WriteString(w, fetch("https://feedx.best/rss/pentitugua.xml", func(s string) string {
+		return strings.ReplaceAll(s, "www.dapenti.com:99", "imgc.1see.org")
+	}))
+}
+
+func zhihuDaily(w http.ResponseWriter, req *http.Request) {
+	io.WriteString(w, fetch("https://feedx.best/rss/zhihudaily.xml", func(s string) string {
+		return filterItem(s, time.Now().Day(), 1)
+	}))
+}
+
+func tjxz(w http.ResponseWriter, r *http.Request) {
+	io.WriteString(w, fetch("https://feedx.best/rss/tjxz.xml", func(s string) string {
+		return filterItem(s, time.Now().Day(), 0)
+	}))
+}
+func bbcLearn(w http.ResponseWriter, _ *http.Request) {
+	io.WriteString(w, fetch("https://www.bbc.co.uk/learningenglish/chinese", func(s string) string {
+		return bbclearn.LearnParse(s, 1)
+	}))
+}
+
+func theNewYorker(w http.ResponseWriter, r *http.Request) {
+	io.WriteString(w, fetch("https://feedx.best/rss/newyorker.xml", func(s string) string {
+		return filterItem(s, time.Now().Day(), 1)
+	}))
+}
+
+func main() {
+	port := os.Getenv("port")
+	if port == "" {
+		port = ":80"
+	}
+	http.HandleFunc("/pentitugua", penti)
+	http.HandleFunc("/zhihuDaily", zhihuDaily)
+	http.HandleFunc("/tjxz", tjxz)
+	http.HandleFunc("/bbcLearn", bbcLearn)
+	http.HandleFunc("/theNewYorker", theNewYorker)
+	log.Fatal(http.ListenAndServe(port, nil))
+}