eh-downloader/internal/eh/eh_test.go

package eh

import "testing"

// ---- 输入解析 ----

func TestParseGalleryInput(t *testing.T) {
	tests := []struct {
		name string
		in   string
		want []GalleryRef
	}{
		{
			name: "标准完整链接",
			in:   "https://e-hentai.org/g/4239272/55e069b6b9",
			want: []GalleryRef{{GID: "4239272", Token: "55e069b6b9"}},
		},
		{
			name: "多行多链接",
			in: "https://e-hentai.org/g/1111111/aaaaaaaaaa\n" +
				"https://e-hentai.org/g/2222222/bbbbbbbbbb",
			want: []GalleryRef{
				{GID: "1111111", Token: "aaaaaaaaaa"},
				{GID: "2222222", Token: "bbbbbbbbbb"},
			},
		},
		{
			name: "竖线分隔一行多个",
			in:   "https://e-hentai.org/g/1111111/aaaaaaaaaa|https://e-hentai.org/g/2222222/bbbbbbbbbb",
			want: []GalleryRef{
				{GID: "1111111", Token: "aaaaaaaaaa"},
				{GID: "2222222", Token: "bbbbbbbbbb"},
			},
		},
		{
			name: "去重",
			in:   "https://e-hentai.org/g/1111111/aaaaaaaaaa\nhttps://e-hentai.org/g/1111111/aaaaaaaaaa",
			want: []GalleryRef{{GID: "1111111", Token: "aaaaaaaaaa"}},
		},
		{
			name: "注释行忽略",
			in:   "# 这是注释\nhttps://e-hentai.org/g/1111111/aaaaaaaaaa",
			want: []GalleryRef{{GID: "1111111", Token: "aaaaaaaaaa"}},
		},
		{
			name: "裸 gid 不匹配完整链接",
			in:   "4239274",
			want: nil,
		},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			got := ParseGalleryInput(tt.in)
			if len(got) != len(tt.want) {
				t.Fatalf("数量不对: got %d want %d (%+v)", len(got), len(tt.want), got)
			}
			for i := range got {
				if got[i].GID != tt.want[i].GID || got[i].Token != tt.want[i].Token {
					t.Errorf("第 %d 项不对: got %s/%s want %s/%s",
						i, got[i].GID, got[i].Token, tt.want[i].GID, tt.want[i].Token)
				}
			}
		})
	}
}

func TestExtractGIDs(t *testing.T) {
	got := ExtractGIDs("看这个 4239274 还有 4239274 重复 和 1234567")
	if len(got) != 2 {
		t.Fatalf("应去重成 2 个，实际 %d 个: %v", len(got), got)
	}
	if got[0] != "4239274" || got[1] != "1234567" {
		t.Errorf("结果不对: %v", got)
	}
}

// ---- hath 直链正则（核心） ----

func TestReHath(t *testing.T) {
	// 从 README 里抄的真实样本
	samples := []string{
		`https://oclzqqk.gyktyewfonyf.hath.network:39656/h/4e9af9f...-44168-707-754-wbp/keystamp=1791448200-8b38ea8f6d;fileindex=270107377;xres=800/NH.webp`,
		`<img src="https://abc.hath.network:8080/h/xyz/gid-1.jpg">`,
		`"https://foo.bar.hath.network/h/a/b/c.png"`,
	}
	for _, s := range samples {
		m := reHath.FindString(s)
		if m == "" {
			t.Errorf("没匹配到: %s", s)
			continue
		}
		if m[len(m)-1] == '"' || m[len(m)-1] == '>' {
			t.Errorf("匹配多了尾部字符: %q", m)
		}
	}

	// 不该匹配的
	for _, s := range []string{
		`https://e-hentai.org/s/abc/123-1`,
		`https://example.com/h/foo`,
	} {
		if m := reHath.FindString(s); m != "" {
			t.Errorf("不该匹配却匹配了: %q", m)
		}
	}
}

func TestApplyXres(t *testing.T) {
	const base = "https://a.hath.network:39656/h/x/keystamp=1;fileindex=2;xres=800/NH.webp"

	cases := []struct {
		xres string
		want string
	}{
		{"auto", base},
		{"", base},
		{"org", "https://a.hath.network:39656/h/x/keystamp=1;fileindex=2;xres=org/NH.webp"},
		{"1920", "https://a.hath.network:39656/h/x/keystamp=1;fileindex=2;xres=1920/NH.webp"},
	}
	for _, c := range cases {
		cl := NewClient(0, "", "off", c.xres)
		if got := cl.applyXres(base); got != c.want {
			t.Errorf("xres=%s: got  %s\n        want %s", c.xres, got, c.want)
		}
	}
}

// ---- 缩略图链接解析（画廊页分页的关键） ----

func TestReThumbLink(t *testing.T) {
	// 模拟 EH 画廊页里的一行
	html := `<a href="/s/bb06b5db77/4239250-1"><div title="Page 1: 001.jpg" class="lazy"></div></a>
	         <a href="https://e-hentai.org/s/cc11ee22ff/4239250-2"><div title="Page 2: 002.png"></div></a>`

	ms := reThumbLink.FindAllStringSubmatch(html, -1)
	if len(ms) != 2 {
		t.Fatalf("应匹配 2 个，实际 %d", len(ms))
	}
	if ms[0][1] != "/s/bb06b5db77/4239250-1" {
		t.Errorf("第 1 个路径不对: %q", ms[0][1])
	}
	if ms[0][2] != "1" || ms[0][3] != "001.jpg" {
		t.Errorf("第 1 个页码/文件名不对: page=%s name=%q", ms[0][2], ms[0][3])
	}
	if ms[1][1] != "https://e-hentai.org/s/cc11ee22ff/4239250-2" {
		t.Errorf("第 2 个路径不对: %q", ms[1][1])
	}
}

func TestReTotal(t *testing.T) {
	for _, s := range []string{
		"Showing 1 - 20 of 206 images",
		"Showing 21 - 40 of 206 images",
	} {
		m := reTotal.FindStringSubmatch(s)
		if m == nil {
			t.Fatalf("没匹配: %s", s)
		}
		if m[1] != "206" {
			t.Errorf("总数不对: %s", m[1])
		}
	}
}

// ---- meta 解析 ----

func TestParseMeta(t *testing.T) {
	html := `<meta name="description" content="Free Hentai Doujinshi Gallery: Something" />
	<div id="taglist">
	  <table>
	    <tr><td class="tc">artist:</td><td><div class="gtl"><a href="/tag/artist:noxe">Noxe</a><a href="/tag/artist:other">Other</a></div></td></tr>
	    <tr><td class="tc">group:</td><td><div class="gtl"><a href="/tag/group:criticalhai">CriticalHAI</a></div></td></tr>
	    <tr><td class="tc">parody:</td><td><div class="gtl"><a href="/tag:parody:wuthering">wuthering waves</a></div></td></tr>
	    <tr><td class="tc">language:</td><td><div class="gtl"><a href="/tag:language:chinese">chinese</a></div></td></tr>
	    <tr><td class="tc">other:</td><td><div class="gtl"><a href="/tag:qingxiao">qingxiao</a><a href="/tag:rover">rover</a><a href="/tag:more">more</a></div></td></tr>
	  </table>
	</div>
	<a href="/uploader/velius11">uploader</a>`

	meta := parseMeta(html, "[Noxe] Title (Circle)")
	if meta.Category != "doujinshi" {
		t.Errorf("分类不对: %q", meta.Category)
	}
	if meta.Artist != "Noxe" {
		t.Errorf("作者不对: %q", meta.Artist)
	}
	if meta.Group != "CriticalHAI" {
		t.Errorf("社团不对: %q", meta.Group)
	}
	if meta.Series != "wuthering waves" {
		t.Errorf("系列不对: %q", meta.Series)
	}
	if meta.Lang != "chinese" {
		t.Errorf("语言不对: %q", meta.Lang)
	}
	// more 按钮要过滤掉
	if len(meta.Tags) != 2 || meta.Tags[0] != "qingxiao" || meta.Tags[1] != "rover" {
		t.Errorf("标签不对(more 未过滤): %v", meta.Tags)
	}
	if meta.Uploader != "velius11" {
		t.Errorf("上传者不对: %q", meta.Uploader)
	}
}

// 标题约定兜底：[作者] 标题 (社团)
func TestParseMetaTitleFallback(t *testing.T) {
	html := `<div id="taglist"></div>`
	meta := parseMeta(html, "[Noxe] 0820 Qingxiao [AI Generated] (CriticalHAI)")
	if meta.Artist != "Noxe" {
		t.Errorf("作者兜底失败: %q", meta.Artist)
	}
	if meta.Group != "CriticalHAI" {
		t.Errorf("社团兜底失败: %q", meta.Group)
	}
}

// ---- HTML 实体 ----

func TestDecodeEntities(t *testing.T) {
	cases := map[string]string{
		"a &amp; b":     "a & b",
		"&lt;tag&gt;":   "<tag>",
		"&quot;x&quot;": `"x"`,
		"&#65;&#66;":    "AB",
		"&#x41;":       "A",
		"plain":        "plain",
		"&#x1F600;":    "\U0001F600",
		"no entities!": "no entities!",
	}
	for in, want := range cases {
		if got := DecodeEntities(in); got != want {
			t.Errorf("DecodeEntities(%q) = %q, want %q", in, got, want)
		}
	}
}