Cyrene_v2/internal/runtime/multimodal/vision/upload_test.go

package vision

import (
	"context"
	"errors"
	"path/filepath"
	"strings"
	"sync/atomic"
	"testing"

	"git.yeij.top/AskaEth/Cyrene_v2/internal/config"
)

// uploadTestName 是一个形状合法的上传文件名（32 位小写十六进制 + .png）。
const uploadTestName = "0123456789abcdef0123456789abcdef.png"

// TestLocalUploadPath 是上传 URL → 本地路径的路径穿越防护用例表：
// 只有严格形状才被接受，任何分隔符/点号段/编码变体都在触碰文件系统前被拒。
func TestLocalUploadPath(t *testing.T) {
	t.Parallel()
	dir := filepath.Join("tmp", "uploads")
	tests := []struct {
		name    string
		rawURL  string
		dir     string
		want    string
		wantHit bool
	}{
		{name: "http upload url", rawURL: "http://192.168.1.5:10942/api/v1/files/" + uploadTestName,
			dir: dir, want: filepath.Join(dir, uploadTestName), wantHit: true},
		{name: "https upload url", rawURL: "https://box.local/api/v1/files/" + uploadTestName,
			dir: dir, want: filepath.Join(dir, uploadTestName), wantHit: true},
		{name: "traversal", rawURL: "http://h/api/v1/files/../../../data/cyrene.db", dir: dir, wantHit: false},
		{name: "encoded traversal", rawURL: "http://h/api/v1/files/%2e%2e%2fcyrene.db", dir: dir, wantHit: false},
		{name: "backslash", rawURL: `http://h/api/v1/files/..\cyrene.db`, dir: dir, wantHit: false},
		{name: "absolute path name", rawURL: "http://h/api/v1/files//etc/passwd", dir: dir, wantHit: false},
		{name: "uppercase hex", rawURL: "http://h/api/v1/files/" + strings.ToUpper(uploadTestName), dir: dir, wantHit: false},
		{name: "short stem", rawURL: "http://h/api/v1/files/abcdef.png", dir: dir, wantHit: false},
		{name: "non hex stem", rawURL: "http://h/api/v1/files/0123456789abcdef0123456789abcdeg.png", dir: dir, wantHit: false},
		{name: "html extension", rawURL: "http://h/api/v1/files/" + strings.TrimSuffix(uploadTestName, ".png") + ".html", dir: dir, wantHit: false},
		{name: "svg extension", rawURL: "http://h/api/v1/files/" + strings.TrimSuffix(uploadTestName, ".png") + ".svg", dir: dir, wantHit: false},
		{name: "extra segment", rawURL: "http://h/api/v1/files/sub/" + uploadTestName, dir: dir, wantHit: false},
		{name: "query string", rawURL: "http://h/api/v1/files/" + uploadTestName + "?x=1", dir: dir, wantHit: false},
		{name: "fragment", rawURL: "http://h/api/v1/files/" + uploadTestName + "#f", dir: dir, wantHit: false},
		{name: "other route", rawURL: "http://h/api/v1/other/" + uploadTestName, dir: dir, wantHit: false},
		{name: "ftp scheme", rawURL: "ftp://h/api/v1/files/" + uploadTestName, dir: dir, wantHit: false},
		{name: "relative url", rawURL: "/api/v1/files/" + uploadTestName, dir: dir, wantHit: false},
		{name: "remote without uploads dir", rawURL: "http://h/api/v1/files/" + uploadTestName, dir: "", wantHit: false},
		{name: "empty name", rawURL: "http://h/api/v1/files/", dir: dir, wantHit: false},
	}
	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			t.Parallel()
			got, ok := localUploadPath(tt.rawURL, tt.dir)
			if ok != tt.wantHit {
				t.Fatalf("localUploadPath(%q, %q) = (%q, %v), want hit=%v", tt.rawURL, tt.dir, got, ok, tt.wantHit)
			}
			if tt.wantHit && got != tt.want {
				t.Fatalf("path = %q, want %q", got, tt.want)
			}
		})
	}
}

// TestOpenAICompatibleAdapter_ReadsOwnUploadLocally 覆盖同源 upload 产物的
// 快路径：URL 指向本服务的 /api/v1/files/<name> 时直接读盘，不发起网络
// 请求（URL 的 host 在此不可达，若走了 HTTP 必然失败/超时）。
func TestOpenAICompatibleAdapter_ReadsOwnUploadLocally(t *testing.T) {
	t.Parallel()
	png := pngBytes(t)
	dataDir := t.TempDir()
	uploads := filepath.Join(dataDir, "uploads")
	if err := writeFileBytesAll(uploads, uploadTestName, png); err != nil {
		t.Fatalf("seed upload: %v", err)
	}
	upSrv, rec := newImageURLRecorderServer(t)
	ad := newAdapter(upSrv.URL, WithUploadsDir(uploads))

	// 127.0.0.1:1 上是没有服务的：一旦退回网络取回就会立刻失败。
	srcURL := "http://127.0.0.1:1/api/v1/files/" + uploadTestName
	if _, err := ad.Describe(context.Background(), ImageRef{Source: SourceURL, Data: srcURL}, ""); err != nil {
		t.Fatalf("Describe via the local upload path: %v", err)
	}
	_, urls := rec.seen()
	if len(urls) != 1 || urls[0] != DataURI("image/png", png) {
		t.Fatalf("image_url = %q, want the local file inlined", urls)
	}
}

// TestOpenAICompatibleAdapter_MissingUploadFailsLoudly 钉住「认得的上传 URL
// 但文件不在」不会偷偷退回网络：错误如实上报（不静默换成别的内容）。
func TestOpenAICompatibleAdapter_MissingUploadFailsLoudly(t *testing.T) {
	t.Parallel()
	uploads := filepath.Join(t.TempDir(), "uploads")
	upSrv, rec := newImageURLRecorderServer(t)
	ad := newAdapter(upSrv.URL, WithUploadsDir(uploads))

	srcURL := "http://127.0.0.1:1/api/v1/files/" + uploadTestName
	_, err := ad.Describe(context.Background(), ImageRef{Source: SourceURL, Data: srcURL}, "")
	if err == nil {
		t.Fatal("Describe succeeded without the upload on disk, want an error")
	}
	if !errors.Is(err, ErrBadImage) {
		t.Fatalf("error = %v, want ErrBadImage", err)
	}
	if calls, _ := rec.seen(); calls != 0 {
		t.Fatalf("upstream was called %d times, want 0", calls)
	}
}

// TestOpenAICompatibleAdapter_TraversalNotReadLocally 是越界文件名的反证
// 用例：本地 secret.png 与远端返回的图内容不同，`../` 名称必须落到远端
// 取回（而不是被读成本地越界文件）。
func TestOpenAICompatibleAdapter_TraversalNotReadLocally(t *testing.T) {
	t.Parallel()
	local := pngBytes(t)
	remote := append([]byte{}, pngBytes(t)...)
	remote = append(remote, 0x00) // 内容不同即可，仍是合法 PNG 前缀
	var hits atomic.Int32

	dataDir := t.TempDir()
	uploads := filepath.Join(dataDir, "uploads")
	if err := writeFileBytesAll(uploads, uploadTestName, local); err != nil {
		t.Fatalf("seed uploads dir: %v", err)
	}
	if err := writeFileBytes(filepath.Join(dataDir, "secret.png"), local); err != nil {
		t.Fatalf("seed secret: %v", err)
	}

	imgSrv := newImageServer(t, remote, "image/png", &hits)
	upSrv, rec := newImageURLRecorderServer(t)
	ad := newAdapter(upSrv.URL, WithUploadsDir(uploads))

	srcURL := imgSrv.URL + "/api/v1/files/../secret.png"
	if _, err := ad.Describe(context.Background(), ImageRef{Source: SourceURL, Data: srcURL}, ""); err != nil {
		t.Fatalf("Describe: %v", err)
	}
	_, urls := rec.seen()
	if len(urls) != 1 {
		t.Fatalf("image_url = %v, want one part", urls)
	}
	if urls[0] == DataURI("image/png", local) {
		t.Fatal("the ../ name was resolved to a local file outside uploads")
	}
	if urls[0] != DataURI("image/png", remote) {
		t.Fatalf("image_url = %q, want the remote payload", urls[0])
	}
	if hits.Load() != 1 {
		t.Fatalf("source fetched %d times, want 1", hits.Load())
	}
}

// TestWithUploadsDirIsOptional 钉住未配置上传目录时行为不变：上传形状的
// URL 仍按普通 URL 取回。
func TestWithUploadsDirIsOptional(t *testing.T) {
	t.Parallel()
	png := pngBytes(t)
	var hits atomic.Int32
	imgSrv := newImageServer(t, png, "image/png", &hits)
	upSrv, rec := newImageURLRecorderServer(t)
	ad := NewOpenAICompatible(config.LLMConfig{BaseURL: upSrv.URL, Model: "vision-test"})

	_, err := ad.Describe(context.Background(), ImageRef{
		Source: SourceURL,
		Data:   imgSrv.URL + "/api/v1/files/" + uploadTestName,
	}, "")
	if err != nil {
		t.Fatalf("Describe: %v", err)
	}
	_, urls := rec.seen()
	if len(urls) != 1 || urls[0] != DataURI("image/png", png) {
		t.Fatalf("image_url = %q, want the fetched payload", urls)
	}
	if hits.Load() != 1 {
		t.Fatalf("source fetched %d times, want 1", hits.Load())
	}
}