MailUI4Agents/gateway/internal/handler/attachments_test.go

package handler

import (
	"strings"
	"testing"
)

// 文件名只用于展示与下载头；磁盘路径由 sha256 派生，
// 因此这里守的是「不污染 HTTP 头、不被当成目录」而非路径穿越。
func TestSanitizeFilename(t *testing.T) {
	cases := []struct{ in, want string }{
		{"report.pdf", "report.pdf"},
		{"中文 文件名.txt", "中文 文件名.txt"},

		// 目录成分必须剥掉（含 Windows 风格）
		{"../../etc/passwd", "passwd"},
		{"/abs/path/x.log", "x.log"},
		{`C:\Users\me\a.txt`, "a.txt"},
		{"a/b/c.txt", "c.txt"},

		// 控制字符会污染 Content-Disposition
		{"bad\r\nname.txt", "badname.txt"},
		{"tab\there.txt", "tabhere.txt"},

		// 无意义的名字兜底
		{"", "unnamed"},
		{"   ", "unnamed"},
		{".", "unnamed"},
		{"..", "unnamed"},
		{"/", "unnamed"},
	}
	for _, c := range cases {
		if got := sanitizeFilename(c.in); got != c.want {
			t.Errorf("sanitizeFilename(%q) = %q, want %q", c.in, got, c.want)
		}
	}
}

// 超长名字按 UTF-8 边界截断，不产生非法序列。
func TestSanitizeFilenameTruncates(t *testing.T) {
	long := strings.Repeat("中", 200) + ".txt" // 每字 3 字节，共 600+
	got := sanitizeFilename(long)
	if len(got) > 255 {
		t.Errorf("截断后 %d 字节，超过 255", len(got))
	}
	for _, r := range got {
		if r == '\uFFFD' {
			t.Fatalf("截断产生非法 UTF-8: %q", got)
		}
	}
}

func TestDetectContentType(t *testing.T) {
	cases := []struct{ declared, filename, want string }{
		{"application/pdf", "x.pdf", "application/pdf"},
		// 客户端没给类型时按扩展名猜
		{"", "notes.txt", "text/plain"},
		{"application/octet-stream", "data.json", "application/json"},
		// 带参数的声明要剥掉参数
		{"text/plain; charset=utf-8", "a.txt", "text/plain"},
		// 认不出就兜底
		{"", "blob.unknownext", "application/octet-stream"},
		{"garbage//not-a-type", "blob.unknownext", "application/octet-stream"},
	}
	for _, c := range cases {
		got := detectContentType(c.declared, c.filename)
		// mime.TypeByExtension 在不同系统上可能返回带参数的值，只比主类型
		if !strings.HasPrefix(got, c.want) {
			t.Errorf("detectContentType(%q, %q) = %q, want prefix %q",
				c.declared, c.filename, got, c.want)
		}
	}
}

// Content-Disposition 必须双写：filename* 承载 UTF-8，filename= 给老客户端兜底。
// 兜底值里的引号/反斜杠/非 ASCII 一律换成下划线，否则能截断响应头。
func TestContentDisposition(t *testing.T) {
	got := contentDisposition("报告 v2.pdf")
	if !strings.HasPrefix(got, "attachment; ") {
		t.Errorf("必须以 attachment 开头: %s", got)
	}
	if !strings.Contains(got, "filename*=UTF-8''") {
		t.Errorf("缺少 RFC 5987 编码: %s", got)
	}
	// 非 ASCII 不能出现在 filename= 的兜底值里
	ascii := got[:strings.Index(got, "filename*=")]
	for _, r := range ascii {
		if r > 0x7e {
			t.Errorf("兜底 filename 含非 ASCII 字符 %q: %s", r, ascii)
		}
	}

	// 引号注入不能逃出引号
	evil := contentDisposition(`a"; x="y`)
	if strings.Contains(evil[:strings.Index(evil, "filename*=")], `"; x=`) {
		t.Errorf("引号未转义，可截断响应头: %s", evil)
	}

	// 控制字符（若绕过 sanitize 直达此处）也不能出现
	ctl := contentDisposition("a\r\nb.txt")
	if strings.ContainsAny(ctl, "\r\n") {
		t.Errorf("响应头含换行: %q", ctl)
	}
}

func TestURLEncodeRFC5987(t *testing.T) {
	cases := []struct{ in, want string }{
		{"abc.txt", "abc.txt"},
		{"a b", "a%20b"},
		{"中", "%E4%B8%AD"},
		{`a"b`, "a%22b"},
	}
	for _, c := range cases {
		if got := urlEncodeRFC5987(c.in); got != c.want {
			t.Errorf("urlEncodeRFC5987(%q) = %q, want %q", c.in, got, c.want)
		}
	}
}

func TestParseAttachmentIDs(t *testing.T) {
	valid := "3f2504e0-4f89-11d3-9a0c-0305e82c3301"

	got, err := parseAttachmentIDs([]string{valid, "  ", ""})
	if err != nil {
		t.Fatalf("合法输入报错: %v", err)
	}
	if len(got) != 1 {
		t.Errorf("空白项应被忽略，得到 %d 个", len(got))
	}

	if _, err := parseAttachmentIDs([]string{"not-a-uuid"}); err == nil {
		t.Error("非法 UUID 应报错")
	}

	if got, err := parseAttachmentIDs(nil); err != nil || len(got) != 0 {
		t.Errorf("nil 应返回空列表，得到 %v, %v", got, err)
	}
}