fix(task): auto-link optional-metadata partial image searches (#368)

2026-10-08 16:54:48 +08:00
parent e450ac2898
commit 15ccab3960
2 changed files with 353 additions and 3 deletions
@@ -2,6 +2,7 @@ package task

 import (
 	"context"
+	"encoding/json"
 	"errors"
 	"fmt"
 	"log"
@@ -15,8 +16,8 @@ import (
 )

 // autoLinkImageSearch is deliberately a narrow post-result hook. It only
-// writes an association when the task was a complete collection, the frozen
-// original association is unchanged, and the target product is active. Manual
+// writes an association when the collection is complete or missing only optional
+// metadata, the frozen original association is unchanged, and the target product is active. Manual
 // changes made while the Agent was running win by the CAS predicate.
 // autoLinkedImageSearch 汇报本次是否确实写入了关联，以及关联覆盖了哪些 SYB 明细。
 // 只有确实写入时才值得触发后续的规格匹配（#287）。
@@ -32,7 +33,7 @@ func (service *Service) autoLinkImageSearch(ctx context.Context, taskID uint64)
 	if err := service.DB.WithContext(ctx).First(&task, taskID).Error; err != nil {
 		return autoLinkedImageSearch{}, internalError(err)
 	}
-	if task.Source != models.CollectionTaskSourceImageSearch || task.Status != models.TaskStatusCompleted || task.PDDProductID == nil || task.ImageSearchSnapshot == nil {
+	if task.Source != models.CollectionTaskSourceImageSearch || !imageSearchResultAllowsAutoLink(task) || task.PDDProductID == nil || task.ImageSearchSnapshot == nil {
 		return autoLinkedImageSearch{}, nil
 	}
 	snapshot, err := loadImageSearchSnapshot(task)
@@ -94,6 +95,31 @@ func (service *Service) autoLinkImageSearch(ctx context.Context, taskID uint64)
 	}, nil
 }

+// #368: Keep completed behavior unchanged. A partial result is eligible only
+// with explicit, nonempty evidence that every missing field is optional.
+// Unknown/malformed missing data must not silently widen this whitelist.
+func imageSearchResultAllowsAutoLink(task models.CollectionTask) bool {
+	if task.Status == models.TaskStatusCompleted {
+		return true
+	}
+	if task.Status != models.TaskStatusCompletedPartial || task.MissingJSON == nil {
+		return false
+	}
+	var missing []string
+	if err := json.Unmarshal([]byte(*task.MissingJSON), &missing); err != nil || len(missing) == 0 {
+		return false
+	}
+	for _, field := range missing {
+		switch field {
+		case "reviewCount", "salesText", "shopName":
+			// Optional descriptive metadata; never colors, sizes or prices.
+		default:
+			return false
+		}
+	}
+	return true
+}
+
 // matchSpecsAfterImageSearch triggers the existing SYB batch spec match for the
 // details the image-search task just linked (#287).
 //
@@ -0,0 +1,324 @@
+package task
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"net/http"
+	"net/http/httptest"
+	"reflect"
+	"sync/atomic"
+	"testing"
+
+	"go-admin/app/goauto/models"
+	"go-admin/app/goauto/shopeeproduct"
+
+	"github.com/google/uuid"
+	"gorm.io/gorm"
+)
+
+// This fixture leaves the hook uncalled so tests can change the submitted
+// status, missing fields and association independently of the assertions.
+func seedPartialImageSearch(t *testing.T, db *gorm.DB, deviceID *uint64) (models.CollectionTask, models.ShopeeProduct) {
+	t.Helper()
+	record := createTask(t, db, deviceID)
+	if err := db.Model(&models.PDDProduct{}).Where("id = ?", *record.PDDProductID).Update("status", "active").Error; err != nil {
+		t.Fatal(err)
+	}
+	shopee := models.ShopeeProduct{
+		ShopeeItemID: "test-item", Title: "test", Currency: "TWD",
+		SpecsJSON: `[{"name":"颜色","role":"color","values":[{"name":"黑色","source":"import"}]},{"name":"尺码","role":"size","values":[{"name":"M","source":"import"}]}]`,
+	}
+	if err := db.Create(&shopee).Error; err != nil {
+		t.Fatal(err)
+	}
+	syb := models.SYBProduct{
+		OrderCode: "TEST-ORDER", DetailID: 1, StockID: 1, ShopeeItemID: shopee.ShopeeItemID,
+		ShopeeProductID: &shopee.ID, TargetColor: "黑色", TargetSize: "M",
+		Quantity: 1, UnitPriceCent: 1000, ParseStatus: "success", RawJSON: "{}",
+	}
+	if err := db.Create(&syb).Error; err != nil {
+		t.Fatal(err)
+	}
+	snapshot := ImageSearchSnapshot{
+		ShopeeProductID: shopee.ID, RepresentativeSYBProductID: syb.ID, SYBProductIDs: []uint64{syb.ID},
+		ReferenceCurrency: "TWD", ReferencePriceCent: 1000, MaxPriceRatio: 3, PriceGuardSkipped: true,
+		ImageSearchImage: ImageSearchImage{ImageURL: "https://example.invalid/a.jpg", MediaType: "image/jpeg", SizeBytes: 1, SHA256: "0123456789012345678901234567890101234567890123456789012345678901"},
+	}
+	raw, err := json.Marshal(snapshot)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if err := db.Model(&record).Updates(map[string]any{"source": models.CollectionTaskSourceImageSearch, "image_search_snapshot": string(raw)}).Error; err != nil {
+		t.Fatal(err)
+	}
+	return record, shopee
+}
+
+func TestAutoLinkImageSearchPartialMissingWhitelist(t *testing.T) {
+	cases := []struct {
+		name, status, missing string
+		nilMissing, want      bool
+	}{
+		{name: "completed_empty", status: models.TaskStatusCompleted, missing: "[]", want: true},
+		{name: "completed_nil", status: models.TaskStatusCompleted, nilMissing: true, want: true},
+		{name: "nil", nilMissing: true},
+		{name: "blank"},
+		{name: "null", missing: "null"},
+		{name: "empty", missing: "[]"},
+		{name: "malformed", missing: `["reviewCount"`},
+		{name: "object", missing: `{ "reviewCount": true }`},
+		{name: "scalar", missing: `"reviewCount"`},
+		{name: "number_entry", missing: `["reviewCount",1]`},
+		{name: "null_entry", missing: `["reviewCount",null]`},
+		{name: "blank_entry", missing: `[""]`},
+		{name: "unknown", missing: `["reviewCount","futureField"]`},
+		{name: "case_mismatch", missing: `["ReviewCount"]`},
+		{name: "whitespace", missing: `[" reviewCount "]`},
+		{name: "color", missing: `["reviewCount","color"]`},
+		{name: "size", missing: `["size"]`},
+		{name: "price", missing: `["price"]`},
+		{name: "color_price", missing: `["reviewCount","price:黑色"]`},
+		{name: "unsupported_dimension", missing: `["unsupportedDimension:款式"]`},
+		{name: "invalid_dimension", missing: `["spec_dimension_invalid:size"]`},
+		{name: "pending", status: models.TaskStatusPending, missing: `["reviewCount"]`},
+		{name: "failed", status: models.TaskStatusFailed, missing: `["reviewCount"]`},
+		{name: "duplicate_optional", missing: `["reviewCount","reviewCount"]`, want: true},
+	}
+	fields := []string{"reviewCount", "salesText", "shopName"}
+	for mask := 1; mask < 8; mask++ {
+		var missing []string
+		for i, field := range fields {
+			if mask&(1<<i) != 0 {
+				missing = append(missing, field)
+			}
+		}
+		raw, _ := json.Marshal(missing)
+		cases = append(cases, struct {
+			name, status, missing string
+			nilMissing, want      bool
+		}{name: fmt.Sprintf("allowed_%d", mask), missing: string(raw), want: true})
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			db := openTaskDatabase(t)
+			record, shopee := seedPartialImageSearch(t, db, nil)
+			status := tc.status
+			if status == "" {
+				status = models.TaskStatusCompletedPartial
+			}
+			var missing *string
+			if !tc.nilMissing {
+				missing = &tc.missing
+			}
+			if err := record.SetStatus(status); err != nil {
+				t.Fatal(err)
+			}
+			if err := db.Model(&record).Updates(map[string]any{"status": status, "missing_json": missing, "active_slot": record.ActiveSlot, "device_run_slot": record.DeviceRunSlot}).Error; err != nil {
+				t.Fatal(err)
+			}
+			linked, err := NewService(db).autoLinkImageSearch(context.Background(), record.ID)
+			if err != nil || linked.Linked != tc.want {
+				t.Fatalf("Linked=%v want=%v error=%v", linked.Linked, tc.want, err)
+			}
+			if err := db.First(&shopee, shopee.ID).Error; err != nil {
+				t.Fatal(err)
+			}
+			if shopee.ImageSearchLinked != tc.want || (shopee.PDDProductID != nil) != tc.want {
+				t.Fatalf("association/marker mismatch: %+v", shopee)
+			}
+			if tc.want && (*shopee.PDDProductID != *record.PDDProductID || len(linked.SYBProductIDs) != 1) {
+				t.Fatal("linked target or downstream SYB identifiers lost")
+			}
+			if err := db.First(&record, record.ID).Error; err != nil {
+				t.Fatal(err)
+			}
+			if record.Status != status || !reflect.DeepEqual(record.MissingJSON, missing) {
+				t.Fatal("eligibility must not rewrite collection status or missing fields")
+			}
+		})
+	}
+}
+
+func TestSubmitImageSearchPartialLinkAndMatch(t *testing.T) {
+	for _, scenario := range []string{"optional_partial", "completed", "critical_missing", "admin_source", "current_page_source", "disabled_pdd", "association_changed", "snapshot_missing", "ai_unavailable", "request_cancelled_during_match"} {
+		t.Run(scenario, func(t *testing.T) {
+			requestContext, cancelRequest := context.WithCancel(context.Background())
+			defer cancelRequest()
+			var syncCalls atomic.Int32
+			erp := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+				syncCalls.Add(1)
+				if scenario == "request_cancelled_during_match" {
+					// Cancel only after the real post-result matching hook has
+					// started. Its detached context must still persist mappings.
+					cancelRequest()
+				}
+				if scenario == "ai_unavailable" {
+					w.WriteHeader(http.StatusServiceUnavailable)
+					return
+				}
+				w.Header().Set("Content-Type", "application/json")
+				fmt.Fprint(w, `{"data":{"color":["黑色"],"size":["M"]}}`)
+			}))
+			defer erp.Close()
+			t.Setenv("GOAUTO_ERPGO_BASE_URL", erp.URL)
+			t.Setenv("GOAUTO_ERPGO_APIKEY", "synthetic-test-key")
+			db := openTaskDatabase(t)
+			device, token := registerTaskDevice(t, db, "partial-image-search")
+			record, shopee := seedPartialImageSearch(t, db, &device.ID)
+			service := newTaskService(db)
+			// Execution is outside this test's scope; submit through the real
+			// authenticated running-owner path with a persisted image-search snapshot.
+			if err := db.Model(&record).Update("status", models.TaskStatusRunning).Error; err != nil {
+				t.Fatal(err)
+			}
+			title, shop, sales, reviews := "商品", "店铺", "已拼100件", int64(12)
+			request := ResultRequest{
+				RequestID: uuid.NewString(), Status: models.TaskStatusCompletedPartial,
+				Product:     ResultProduct{PDDGoodsID: record.GoodsIDSnapshot, Title: &title, ShopName: &shop, SalesText: &sales},
+				Dimensions:  []ResultDimension{{Key: "color", Name: "颜色", Values: []string{"黑色"}}, {Key: "size", Name: "尺码", Values: []string{"M"}}},
+				ColorPrices: []ResultColorPrice{{Color: "黑色", PriceCent: 1000}},
+				SKUs:        []ResultSKU{{Specs: map[string]string{"color": "黑色", "size": "M"}, PriceCent: 1000, Available: true}},
+				Missing:     []string{"reviewCount"},
+			}
+			wantLinked, wantMapped := true, true
+			var unchangedPDD *uint64
+			switch scenario {
+			case "completed":
+				request.Status, request.Missing, request.Product.ReviewCount = models.TaskStatusCompleted, []string{}, &reviews
+			case "critical_missing":
+				request.Missing = append(request.Missing, "size")
+				wantLinked, wantMapped = false, false
+			case "admin_source", "current_page_source":
+				source := models.CollectionTaskSourceAdmin
+				if scenario == "current_page_source" {
+					source = models.CollectionTaskSourceAgentCurrentPage
+				}
+				if err := db.Model(&record).Update("source", source).Error; err != nil {
+					t.Fatal(err)
+				}
+				wantLinked, wantMapped = false, false
+			case "disabled_pdd":
+				if err := db.Model(&models.PDDProduct{}).Where("id = ?", *record.PDDProductID).Update("status", "disabled").Error; err != nil {
+					t.Fatal(err)
+				}
+				wantLinked, wantMapped = false, false
+			case "association_changed":
+				manual := seedAutoLinkPDD(t, db, "manual-pdd")
+				unchangedPDD = &manual.ID
+				if err := db.Model(&shopee).Update("pdd_product_id", manual.ID).Error; err != nil {
+					t.Fatal(err)
+				}
+				wantLinked, wantMapped = false, false
+			case "snapshot_missing":
+				if err := db.Model(&record).Update("image_search_snapshot", nil).Error; err != nil {
+					t.Fatal(err)
+				}
+				wantLinked, wantMapped = false, false
+			case "ai_unavailable":
+				// No exact match and no configured AI provider: the existing
+				// matching path must not invent mappings or undo successful collection.
+				request.Dimensions[0].Values = []string{"白色"}
+				request.Dimensions[1].Values = []string{"L"}
+				request.ColorPrices[0].Color = "白色"
+				request.SKUs[0].Specs = map[string]string{"color": "白色", "size": "L"}
+				wantMapped = false
+			}
+			for attempt := 0; attempt < 2; attempt++ {
+				ctx := context.Background()
+				if attempt == 0 {
+					ctx = requestContext
+				}
+				detail, err := service.SubmitResult(ctx, record.ID, request, token)
+				if scenario == "request_cancelled_during_match" && attempt == 0 {
+					// Detail still uses the original request context. Preserve that
+					// existing return behavior, then inspect the committed state
+					// independently; do not pretend the disconnected client got OK.
+					if requestContext.Err() != context.Canceled || err == nil {
+						t.Fatalf("expected cancelled response read, got %v", err)
+					}
+					detail, err = service.Detail(context.Background(), record.ID)
+				}
+				if err != nil {
+					t.Fatalf("submit attempt %d: %v", attempt, err)
+				}
+				if detail.Task.Status != request.Status || !reflect.DeepEqual(detail.Missing, request.Missing) || detail.Replayed != (attempt == 1) {
+					t.Fatalf("submission/replay changed result: status=%s missing=%v replayed=%v", detail.Task.Status, detail.Missing, detail.Replayed)
+				}
+				if len(detail.SKUs) != 1 || len(detail.Dimensions) != 2 {
+					t.Fatal("result children missing or duplicated")
+				}
+				if err := db.First(&shopee, shopee.ID).Error; err != nil {
+					t.Fatal(err)
+				}
+				if shopee.ImageSearchLinked != wantLinked {
+					t.Fatalf("image search marker=%v", shopee.ImageSearchLinked)
+				}
+				wantPDD := unchangedPDD
+				if wantLinked {
+					wantPDD = record.PDDProductID
+				}
+				if !reflect.DeepEqual(shopee.PDDProductID, wantPDD) {
+					t.Fatal("wrong association or concurrent change overwritten")
+				}
+				wantCalls := int32(0)
+				if wantLinked {
+					wantCalls = 1
+				}
+				if syncCalls.Load() != wantCalls {
+					t.Fatalf("matching hook calls=%d want=%d", syncCalls.Load(), wantCalls)
+				}
+				specs, err := shopeeproduct.Unmarshal(shopee.SpecsJSON)
+				if err != nil {
+					t.Fatal(err)
+				}
+				if len(specs) != 2 {
+					t.Fatalf("unexpected archive dimensions: %d", len(specs))
+				}
+				for _, dimension := range specs {
+					for _, value := range dimension.Values {
+						if wantMapped {
+							if value.Mapping == nil || value.Mapping.Status != shopeeproduct.MappingStatusConfirmed || value.Mapping.PDDValue != value.Name {
+								t.Fatalf("matching not persisted: %+v", value)
+							}
+						} else if value.Mapping != nil {
+							t.Fatalf("unexpected mapping: %+v", value)
+						}
+					}
+				}
+			}
+			if scenario == "optional_partial" {
+				// A second result for the same image carries the same original
+				// association snapshot. The first committed link wins; no fallback
+				// to another result or a second matching pass is introduced.
+				if err := db.First(&record, record.ID).Error; err != nil {
+					t.Fatal(err)
+				}
+				second := createTask(t, db, &device.ID)
+				if err := db.Model(&second).Updates(map[string]any{
+					"source": models.CollectionTaskSourceImageSearch, "image_search_snapshot": record.ImageSearchSnapshot,
+					"status": models.TaskStatusRunning,
+				}).Error; err != nil {
+					t.Fatal(err)
+				}
+				request.RequestID, request.Product.PDDGoodsID = uuid.NewString(), second.GoodsIDSnapshot
+				if _, err := service.SubmitResult(context.Background(), second.ID, request, token); err != nil {
+					t.Fatal(err)
+				}
+				if err := db.First(&shopee, shopee.ID).Error; err != nil {
+					t.Fatal(err)
+				}
+				if !reflect.DeepEqual(shopee.PDDProductID, record.PDDProductID) || syncCalls.Load() != 1 {
+					t.Fatal("competing image-search result replaced association or retriggered matching")
+				}
+			}
+			var purchases int64
+			if err := db.Model(&models.PurchaseTask{}).Count(&purchases).Error; err != nil {
+				t.Fatal(err)
+			}
+			if purchases != 0 {
+				t.Fatal("matching must not create a purchase task")
+			}
+		})
+	}
+}