mirror of
https://github.com/truewhile/MeBox.git
synced 2026-10-06 05:16:38 +08:00
修复段评往下滑一直闪：评论接口分页改游标，页面还在按 page 翻页
上游评论页按 page=N 翻页，但列表接口 /para_review 早已改成游标分页（响应给
next_cursor，请求要带 cursor=）。接口不认识 page，于是每次「加载更多」都原样
返回第一页：一条新评论都加不进去，has_more 又始终是 true，评论页便一遍遍触发
加载更多、反复闪骨架屏——用户看到的就是段评往下滑一闪一闪，评论一条没多。
评论页是上游 HTML 改不了，于是在承载页面的接口代理里做最小适配：
- 请求带 page=N（N>1）且有上一页给的游标时，换成 cursor= 请求；
- 某页返回的 comment_id 全是已经见过的，就把 has_more 改成 false 收尾，
避免页面在「加载更多 → 没有新内容」之间空转。
只认路径以 /para_review 结尾、带 page 且不带 cursor 的请求，其余原样透传。
This commit is contained in:
truewhile
@@ -0,0 +1,187 @@
|
||||
package reader
|
||||
|
||||
import (
|
||||
"encoding/json"
|
||||
"io"
|
||||
"net/http"
|
||||
"net/http/httptest"
|
||||
"strings"
|
||||
"sync"
|
||||
"testing"
|
||||
|
||||
"github.com/truewhile/MeBox/internal/service/reader/rule"
|
||||
)
|
||||
|
||||
// 本文件：段落评论翻页适配（browser_comment_page.go）的测试。
|
||||
//
|
||||
// 覆盖两件事：
|
||||
// 1. page=N（N>1）被换成上一页响应给的 cursor；
|
||||
// 2. 整页重复（接口不认 page 时的表现）时把 has_more 改成 false，页面才会收尾。
|
||||
|
||||
const commentParaBase = "https://cmt.example.com/para_review?item_id=X¶=1&source=QQ"
|
||||
|
||||
func TestCommentPagerRewritesPageToCursor(t *testing.T) {
|
||||
p := newCommentPager()
|
||||
|
||||
// 第一页：原样请求，响应里的游标被记下
|
||||
target, key := p.rewriteRequest(commentParaBase + "&page=1")
|
||||
if key == "" {
|
||||
t.Fatal("段落评论接口应被识别")
|
||||
}
|
||||
if target != commentParaBase+"&page=1" {
|
||||
t.Fatalf("第一页不该改写: %q", target)
|
||||
}
|
||||
p.observeResponse(key, `{"code":0,"data":{"comments":[{"comment_id":"c1"}],"has_more":true,"next_cursor":"1500000000000"}}`)
|
||||
|
||||
// 第二页：改成带 cursor 的请求，page 不再出现
|
||||
target, key2 := p.rewriteRequest(commentParaBase + "&page=2")
|
||||
if key2 != key {
|
||||
t.Fatalf("同一段落的状态键应一致: %q vs %q", key2, key)
|
||||
}
|
||||
if !strings.Contains(target, "cursor=1500000000000") || strings.Contains(target, "page=") {
|
||||
t.Fatalf("第二页应换成 cursor 请求: %q", target)
|
||||
}
|
||||
|
||||
// 拿到新评论：游标前进到下一页
|
||||
p.observeResponse(key, `{"code":0,"data":{"comments":[{"comment_id":"c2"}],"has_more":true,"next_cursor":"1400000000000"}}`)
|
||||
target, _ = p.rewriteRequest(commentParaBase + "&page=3")
|
||||
if !strings.Contains(target, "cursor=1400000000000") {
|
||||
t.Fatalf("第三页应使用新的游标: %q", target)
|
||||
}
|
||||
|
||||
// 重新加载（page=1）会把整轮状态清掉：没有游标就原样透传
|
||||
p.rewriteRequest(commentParaBase + "&page=1")
|
||||
target, _ = p.rewriteRequest(commentParaBase + "&page=2")
|
||||
if strings.Contains(target, "cursor=") {
|
||||
t.Fatalf("重置后不该还带着旧游标: %q", target)
|
||||
}
|
||||
}
|
||||
|
||||
func TestCommentPagerStopsOnRepeatedPage(t *testing.T) {
|
||||
p := newCommentPager()
|
||||
_, key := p.rewriteRequest(commentParaBase + "&page=1")
|
||||
first := `{"code":0,"data":{"comments":[{"comment_id":"c1","like_count":562,"user":{"user_id":601948729051219}}],"has_more":true,"next_cursor":"1500000000000"}}`
|
||||
if got := p.observeResponse(key, first); got != first {
|
||||
t.Fatalf("第一页不该被改写: %q", got)
|
||||
}
|
||||
|
||||
// 同一页又被返回一次（接口忽略 page）：改 has_more，让页面显示「没有更多了」
|
||||
repeat := `{"code":0,"data":{"comments":[{"comment_id":"c1","like_count":562,"user":{"user_id":601948729051219}}],"has_more":true,"next_cursor":"1500000000000"}}`
|
||||
got := p.observeResponse(key, repeat)
|
||||
if !strings.Contains(got, `"has_more":false`) {
|
||||
t.Fatalf("重复页应把 has_more 改成 false: %q", got)
|
||||
}
|
||||
// 只动 has_more 一处，其它字段（尤其是大整数）保持原样
|
||||
if strings.Replace(got, `"has_more":false`, `"has_more":true`, 1) != first {
|
||||
t.Fatalf("除 has_more 外不应改动响应: %q", got)
|
||||
}
|
||||
// 重复页不再推进游标
|
||||
target, _ := p.rewriteRequest(commentParaBase + "&page=2")
|
||||
if !strings.Contains(target, "cursor=1500000000000") {
|
||||
t.Fatalf("游标不该被重复页顶掉: %q", target)
|
||||
}
|
||||
}
|
||||
|
||||
func TestCommentPagerIgnoresOtherRequests(t *testing.T) {
|
||||
p := newCommentPager()
|
||||
cases := []string{
|
||||
"https://cmt.example.com/book/detail?page=2", // 不是评论接口
|
||||
"https://cmt.example.com/para_review?item_id=X", // 没有 page
|
||||
"https://cmt.example.com/para_review?item_id=X&cursor=1", // 已经在用游标
|
||||
"https://cmt.example.com/para_review?item_id=X&page=0", // page 非法
|
||||
"https://cmt.example.com/para_review?item_id=X&sort_by=1&", // 没有 page
|
||||
}
|
||||
for _, target := range cases {
|
||||
got, key := p.rewriteRequest(target)
|
||||
if got != target || key != "" {
|
||||
t.Fatalf("不该改写的请求被处理了: %q -> %q (key=%q)", target, got, key)
|
||||
}
|
||||
}
|
||||
// 非 JSON 响应原样返回
|
||||
if got := p.observeResponse("k", "<html></html>"); got != "<html></html>" {
|
||||
t.Fatalf("非 JSON 响应被改动: %q", got)
|
||||
}
|
||||
}
|
||||
|
||||
// TestBrowserXHRCommentPagerFollowsCursor 端到端：页面按 page 翻页，
|
||||
// 代理层换成 cursor，接口给出新评论，最后一页重复时收尾。
|
||||
func TestBrowserXHRCommentPagerFollowsCursor(t *testing.T) {
|
||||
svc, _ := newLoginTestService(t)
|
||||
|
||||
var mu sync.Mutex
|
||||
seenCursors := []string{}
|
||||
seenPages := []string{}
|
||||
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
|
||||
switch r.URL.Path {
|
||||
case "/page":
|
||||
w.Header().Set("Content-Type", "text/html; charset=utf-8")
|
||||
_, _ = io.WriteString(w, `<html><body>comment page</body></html>`)
|
||||
case "/para_review":
|
||||
q := r.URL.Query()
|
||||
mu.Lock()
|
||||
seenCursors = append(seenCursors, q.Get("cursor"))
|
||||
seenPages = append(seenPages, q.Get("page"))
|
||||
mu.Unlock()
|
||||
w.Header().Set("Content-Type", "application/json; charset=utf-8")
|
||||
// 上游只认 cursor：不认识 page，靠 next_cursor 一页页往前。
|
||||
switch q.Get("cursor") {
|
||||
case "":
|
||||
_, _ = io.WriteString(w, `{"code":0,"data":{"comments":[{"comment_id":"c1"}],"has_more":true,"next_cursor":"1500000000000"}}`)
|
||||
case "1500000000000":
|
||||
_, _ = io.WriteString(w, `{"code":0,"data":{"comments":[{"comment_id":"c2"}],"has_more":true,"next_cursor":"1400000000000"}}`)
|
||||
case "1400000000000":
|
||||
_, _ = io.WriteString(w, `{"code":0,"data":{"comments":[{"comment_id":"c3"}],"has_more":false,"next_cursor":""}}`)
|
||||
default:
|
||||
// 兜底分支：上游把同一页又给了一遍（页面的 page 循环就会走到这里）
|
||||
_, _ = io.WriteString(w, `{"code":0,"data":{"comments":[{"comment_id":"c3"}],"has_more":true,"next_cursor":"1300000000000"}}`)
|
||||
}
|
||||
default:
|
||||
http.NotFound(w, r)
|
||||
}
|
||||
}))
|
||||
defer srv.Close()
|
||||
|
||||
entry, err := svc.registerBrowser(t.Context(), srv.URL, "src-1", readerTestUserID,
|
||||
browserCookieTarget{}, rule.BrowserTask{URL: srv.URL + "/page", Title: "段评"}, browserModeOpen)
|
||||
if err != nil {
|
||||
t.Fatalf("登记承载页面失败: %v", err)
|
||||
}
|
||||
call := func(page string) string {
|
||||
t.Helper()
|
||||
res, err := svc.ProxyBrowserXHR(t.Context(), entry.id, http.MethodGet,
|
||||
srv.URL+"/para_review?item_id=X¶=1&source=QQ&page="+page, nil, "")
|
||||
if err != nil {
|
||||
t.Fatalf("代理请求失败: %v", err)
|
||||
}
|
||||
return res.Body
|
||||
}
|
||||
|
||||
if body := call("1"); !strings.Contains(body, `"has_more":true`) || !strings.Contains(body, "c1") {
|
||||
t.Fatalf("第一页响应异常: %q", body)
|
||||
}
|
||||
if body := call("2"); !strings.Contains(body, "c2") {
|
||||
t.Fatalf("第二页应拿到下一页评论: %q", body)
|
||||
}
|
||||
if body := call("3"); !strings.Contains(body, "c3") || !strings.Contains(body, `"has_more":false`) {
|
||||
t.Fatalf("第三页应是真正的最后一页: %q", body)
|
||||
}
|
||||
// 页面不知道已经到底，还可能再要一页：上游又把最后一页给了一遍，
|
||||
// 适配层要把 has_more 改成 false，页面才会显示「没有更多了」并停下。
|
||||
last := call("4")
|
||||
if !strings.Contains(last, `"has_more":false`) {
|
||||
t.Fatalf("整页重复时应收尾（has_more=false）: %q", last)
|
||||
}
|
||||
|
||||
mu.Lock()
|
||||
defer mu.Unlock()
|
||||
if strings.Join(seenPages, ",") != "1,,,4" {
|
||||
t.Fatalf("page 参数只应出现在没有游标可用的请求上: %v", seenPages)
|
||||
}
|
||||
if strings.Join(seenCursors, ",") != ",1500000000000,1400000000000," {
|
||||
t.Fatalf("游标传递不符预期: %v", seenCursors)
|
||||
}
|
||||
var env map[string]any
|
||||
if err := json.Unmarshal([]byte(last), &env); err != nil {
|
||||
t.Fatalf("收尾后的响应不是合法 JSON: %v", err)
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user