lunar-mini/server/internal/service/wiki_sync_test.go

package service

import (
	"strings"
	"testing"
)

// 取自百度百科 /cms/home/eventsOnHistory/08.json 的真实片段（8月12日/13日）
const sampleBaikeJSON = `{"08":{
  "0812":[
    {"year":"-30","title":"埃及艳后<a target=\"_blank\" href=\"https://baike.baidu.com/item/x\">克利奥帕特拉七世</a>逝世","festival":"","link":"https://baike.baidu.com/item/x","type":"death","desc":"克利奥帕特拉七世"},
    {"year":"1848","title":"英国著名小说家<a target=\"_blank\" href=\"x\">乔治·艾略特</a>出生","festival":"","link":"x","type":"birth","desc":"x"},
    {"year":"1848","title":"英国著名小说家乔治·艾略特出生","festival":"","link":"x","type":"birth","desc":"x"},
    {"year":"1905","title":"挪威音乐家<a href=\"x\">某</a>指挥首演","festival":"","link":"x","type":"event","desc":"x"}
  ],
  "0813":[
    {"year":"604","title":"隋文帝<a href=\"x\">杨坚</a>逝世","festival":"国际左撇子日","link":"x","type":"death","desc":"x"},
    {"year":"1926","title":"古巴革命领导人<a href=\"x\">卡斯特罗</a>出生","festival":"国际左撇子日","link":"x","type":"birth","desc":"x"},
    {"year":"1910","title":"某事件<a href=\"x\">发生</a>","festival":"国际左撇子日","link":"x","type":"event","desc":"x"},
    {"year":"abc","title":"年份非法的<a href=\"x\">事件</a>条目","festival":"国际左撇子日","link":"x","type":"event","desc":"x"}
  ]
}}`

func TestParseBaikeMonth(t *testing.T) {
	items, err := parseBaikeMonth([]byte(sampleBaikeJSON), 8)
	if err != nil {
		t.Fatalf("parseBaikeMonth 失败: %v", err)
	}

	count := func(kind string) int {
		n := 0
		for _, it := range items {
			if it.Kind == kind {
				n++
			}
		}
		return n
	}

	// 0812: 1 death + 2 birth(重复去重后1) + 1 event = 3；0813: 1 death + 1 birth + 2 event = 4
	if got := count("death"); got != 2 {
		t.Errorf("death 条目数 = %d, 期望 2", got)
	}
	if got := count("birth"); got != 2 { // 0812 重复出生条目应去重
		t.Errorf("birth 条目数 = %d, 期望 2", got)
	}
	if got := count("event"); got != 3 {
		t.Errorf("event 条目数 = %d, 期望 3", got)
	}
	// 国际左撇子日只在 0813，去重后 1 条 festival
	if got := count("festival"); got != 1 {
		t.Errorf("festival 条目数 = %d, 期望 1", got)
	}

	// 校验具体条目
	assertItem := func(kind string, day, wantYear int, wantContent string) {
		for _, it := range items {
			if it.Kind == kind && it.Day == day && it.Content == wantContent {
				if it.Year != wantYear {
					t.Errorf("%q 年份 = %d, 期望 %d", wantContent, it.Year, wantYear)
				}
				if it.Month != 8 {
					t.Errorf("%q 月份 = %d, 期望 8", wantContent, it.Month)
				}
				return
			}
		}
		t.Errorf("未找到条目: kind=%s day=%d content=%q", kind, day, wantContent)
	}
	assertItem("death", 12, -30, "埃及艳后克利奥帕特拉七世逝世")
	assertItem("birth", 12, 1848, "英国著名小说家乔治·艾略特出生")
	assertItem("event", 13, 0, "年份非法的事件条目") // 非法年份应为 0
	assertItem("festival", 13, 0, "国际左撇子日")

	// 不应残留 HTML 标签
	for _, it := range items {
		if strings.Contains(it.Content, "<a") || strings.Contains(it.Content, "href") ||
			strings.Contains(it.Content, "</a>") {
			t.Errorf("条目残留 HTML 标签: %q", it.Content)
		}
	}
}

func TestParseBaikeYear(t *testing.T) {
	cases := map[string]int{
		"1912": 1912,
		"-30":  -30,
		"0":    0,
		"abc":  0,
		"":     0,
		" 77 ": 77,
	}
	for in, want := range cases {
		if got := parseBaikeYear(in); got != want {
			t.Errorf("parseBaikeYear(%q) = %d, 期望 %d", in, got, want)
		}
	}
}

func TestNormalizeKind(t *testing.T) {
	if normalizeKind("event") != "event" || normalizeKind("birth") != "birth" ||
		normalizeKind("death") != "death" || normalizeKind("other") != "" ||
		normalizeKind(" EVENT ") != "event" {
		t.Error("normalizeKind 映射异常")
	}
}

func TestCleanHTMLText(t *testing.T) {
	in := `埃及艳后<a target="_blank" href="x">克利奥帕特拉七世</a>  逝世&nbsp;test`
	want := "埃及艳后克利奥帕特拉七世 逝世 test"
	if got := cleanHTMLText(in); got != want {
		t.Errorf("cleanHTMLText = %q, 期望 %q", got, want)
	}
}