fix(syb): resolve uncertain specs against existing candidates #251

server/app/goauto/sybimport/ai_parse_batch_test.go

@@ -44,6 +44,30 @@ func seedAIParseCandidate(t *testing.T, db *gorm.DB, serverURL string, detailID
 	return applied.SYBProduct
 }
 
+func TestScheduledParseUsesUniqueCandidatesBeforeCallingAI(t *testing.T) {
+	var calls atomic.Int32
+	provider := httptest.NewServer(http.HandlerFunc(func(response http.ResponseWriter, request *http.Request) {
+		calls.Add(1)
+		response.WriteHeader(http.StatusInternalServerError)
+	}))
+	defer provider.Close()
+	db := openTestDB(t)
+	record := seedAIParseCandidate(t, db, provider.URL, 25102, "黑色 XL")
+	run, _, err := sybimport.StartSpecAIParseRun(context.Background(), db, uuid.NewString(), 20)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if err := sybimport.ProcessSpecAIParseRun(context.Background(), db, run.ID); err != nil {
+		t.Fatal(err)
+	}
+	if err := db.First(&record, record.ID).Error; err != nil {
+		t.Fatal(err)
+	}
+	if record.ParseStatus != models.SYBParseStatusSuccess || record.TargetColor != "黑色" || record.TargetSize != "XL" || record.AIConfirmed || calls.Load() != 0 {
+		t.Fatalf("unique candidate must resolve deterministically, status=%s ai=%v calls=%d", record.ParseStatus, record.AIConfirmed, calls.Load())
+	}
+}
+
 func TestScheduledAIParseConfirmsClosedCandidatesAndDoesNotRepeat(t *testing.T) {
 	var calls atomic.Int32
 	provider := httptest.NewServer(http.HandlerFunc(func(response http.ResponseWriter, request *http.Request) {
@@ -53,7 +77,7 @@ func TestScheduledAIParseConfirmsClosedCandidatesAndDoesNotRepeat(t *testing.T)
 	}))
 	defer provider.Close()
 	db := openTestDB(t)
-	record := seedAIParseCandidate(t, db, provider.URL, 19801, "黑色 XL")
+	record := seedAIParseCandidate(t, db, provider.URL, 19801, "黑 XL")
 	rawBefore := record.RawJSON
 	run, created, err := sybimport.StartSpecAIParseRun(context.Background(), db, uuid.NewString(), 20)
 	if err != nil || !created {
@@ -98,7 +122,7 @@ func TestScheduledAIParseLeavesLowConfidenceUnmatchedForSameFingerprint(t *testi
 	}))
 	defer provider.Close()
 	db := openTestDB(t)
-	record := seedAIParseCandidate(t, db, provider.URL, 19802, "黑色 XL")
+	record := seedAIParseCandidate(t, db, provider.URL, 19802, "黑 XL")
 	for i := 0; i < 2; i++ {
 		run, created, err := sybimport.StartSpecAIParseRun(context.Background(), db, uuid.NewString(), 20)
 		if err != nil || !created {
@@ -175,7 +199,7 @@ func TestScheduledAIParseRetriesProviderFailureAtMostThreeTimes(t *testing.T) {
 	}))
 	defer provider.Close()
 	db := openTestDB(t)
-	record := seedAIParseCandidate(t, db, provider.URL, 19805, "黑色 XL")
+	record := seedAIParseCandidate(t, db, provider.URL, 19805, "黑 XL")
 	for attempt := 1; attempt <= 4; attempt++ {
 		run, created, err := sybimport.StartSpecAIParseRun(context.Background(), db, uuid.NewString(), 20)
 		if err != nil || !created {

server/app/goauto/sybimport/apply.go

@@ -96,6 +96,9 @@ func ApplyDetail(ctx context.Context, db *gorm.DB, order OrderInput, detail Deta
 		if err != nil {
 			return err
 		}
+		if shopeeProduct != nil {
+			parsed = ParseWithCandidates(detail.ProductSpec, shopeeProduct.SpecsJSON)
+		}
 		if parsed.Status == models.SYBParseStatusSuccess && shopeeProduct != nil {
 			if err := mergeParsedSpec(tx, shopeeProduct.ID, parsed); err != nil {
 				return err
@@ -271,6 +274,9 @@ func fillEmptyArchiveFields(tx *gorm.DB, product *models.ShopeeProduct, detail D
 // review but are never pushed into the shared archive, so a messy raw string
 // never pollutes the spec list other screens read from.
 func mergeParsedSpec(tx *gorm.DB, shopeeProductID uint64, parsed ParseResult) error {
+	if parsed.matchedCandidates {
+		return nil
+	}
 	var product models.ShopeeProduct
 	if err := tx.First(&product, shopeeProductID).Error; err != nil {
 		return err

server/app/goauto/sybimport/parse.go

@@ -34,6 +34,8 @@ type ParseResult struct {
 	Size   string
 	Status string
 	Note   string
+	// Already present in the archive; do not re-merge or modify mappings.
+	matchedCandidates bool
 }
 
 // Parse implements the #41/#216 rule: split on the last comma, strip 【...】

server/app/goauto/sybimport/parse_candidates.go

@@ -0,0 +1,69 @@
+package sybimport
+
+import (
+	"strings"
+
+	"go-admin/app/goauto/models"
+	"go-admin/app/goauto/shopeeproduct"
+)
+
+// ParseWithCandidates only rescues uncertain parses against previously stored
+// Shopee labels. It never uses the current unconfirmed input as its own evidence.
+func ParseWithCandidates(raw, specsJSON string) ParseResult {
+	baseline := Parse(raw)
+	if baseline.Status != models.SYBParseStatusUncertain {
+		return baseline
+	}
+	specs, err := shopeeproduct.Unmarshal(specsJSON)
+	if err != nil || shopeeproduct.Validate(specs) != nil {
+		return baseline
+	}
+	for _, dimension := range specs {
+		if dimension.Role != shopeeproduct.RoleColor && dimension.Role != shopeeproduct.RoleSize && len(dimension.Values) > 0 {
+			return baseline
+		}
+	}
+	colors, sizes, ambiguous := closedShopeeCandidates(specs)
+	if ambiguous || len(colors)+len(sizes) == 0 {
+		return baseline
+	}
+	if len(colors) == 0 {
+		colors = []string{""}
+	}
+	if len(sizes) == 0 {
+		sizes = []string{""}
+	}
+	input := normalizeCandidateSpec(stripBrackets(raw))
+	type pair struct{ color, size string }
+	matches := map[pair]bool{}
+	for _, color := range colors {
+		for _, size := range sizes {
+			variants := []string{color + "," + size, size + "," + color, color + " " + size, size + " " + color}
+			if color == "" || size == "" {
+				variants = []string{color + size}
+			}
+			for _, variant := range variants {
+				if input == normalizeCandidateSpec(variant) {
+					matches[pair{color, size}] = true
+				}
+			}
+		}
+	}
+	if len(matches) != 1 {
+		return baseline
+	}
+	for match := range matches {
+		return ParseResult{Color: match.color, Size: match.size, Status: models.SYBParseStatusSuccess,
+			Note: "已按蝦皮既有规格候选唯一核对", matchedCandidates: true}
+	}
+	return baseline
+}
+
+func normalizeCandidateSpec(value string) string {
+	value = strings.ReplaceAll(value, "，", ",")
+	parts := strings.Split(value, ",")
+	for i, part := range parts {
+		parts[i] = strings.Join(strings.Fields(part), " ")
+	}
+	return strings.Join(parts, ",")
+}

server/app/goauto/sybimport/parse_candidates_test.go

@@ -0,0 +1,121 @@
+package sybimport_test
+
+import (
+	"context"
+	"encoding/json"
+	"testing"
+
+	"go-admin/app/goauto/models"
+	"go-admin/app/goauto/shopeeproduct"
+	"go-admin/app/goauto/sybimport"
+)
+
+func candidateSpecs(colors, sizes []string) string {
+	dimensions := []shopeeproduct.SpecDimension{}
+	for _, input := range []struct {
+		role   string
+		values []string
+	}{{"color", colors}, {"size", sizes}} {
+		if len(input.values) == 0 {
+			continue
+		}
+		dimension := shopeeproduct.SpecDimension{Name: input.role, Role: input.role}
+		for _, value := range input.values {
+			dimension.Values = append(dimension.Values, shopeeproduct.SpecValue{Name: value, Source: shopeeproduct.ValueSourceImport})
+		}
+		dimensions = append(dimensions, dimension)
+	}
+	encoded, _ := json.Marshal(dimensions)
+	return string(encoded)
+}
+
+func TestCandidateParseRescuesOnlyUniqueWholeInput(t *testing.T) {
+	for _, test := range []struct {
+		name, raw     string
+		colors, sizes []string
+		color, size   string
+	}{
+		{"combined color", "黑色+白色,L", []string{"黑色+白色", "黑色"}, []string{"L", "XL"}, "黑色+白色", "L"},
+		{"spaces in color", "浅  灰色,XL", []string{"浅 灰色"}, []string{"XL"}, "浅 灰色", "XL"},
+		{"full width comma", "黑色，XL", []string{"黑色"}, []string{"XL"}, "黑色", "XL"},
+		{"reverse full width", "XL，黑色", []string{"黑色"}, []string{"XL"}, "黑色", "XL"},
+		{"space separator", "黑色 XL", []string{"黑色"}, []string{"XL"}, "黑色", "XL"},
+		{"size only", "均碼", nil, []string{"均碼"}, "", "均碼"},
+		{"color only", "黑色+白色", []string{"黑色+白色"}, nil, "黑色+白色", ""},
+		{"weight description exact", "黑色，XL 建議55.5-60.0公斤穿", []string{"黑色"}, []string{"XL 建議55.5-60.0公斤穿"}, "黑色", "XL 建議55.5-60.0公斤穿"},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			got := sybimport.ParseWithCandidates(test.raw, candidateSpecs(test.colors, test.sizes))
+			if got.Status != models.SYBParseStatusSuccess || got.Color != test.color || got.Size != test.size {
+				t.Fatalf("unexpected parse: %+v", got)
+			}
+		})
+	}
+}
+
+func TestCandidateParsePreservesAmbiguityAndNormalResults(t *testing.T) {
+	for _, test := range []struct{ raw, specs string }{
+		{"", candidateSpecs([]string{"黑色"}, []string{"XL"})},
+		{"黑色 XL", "broken"},
+		{"黑色 XL", "[]"},
+		{"XL", candidateSpecs([]string{"黑色"}, []string{"XL"})},
+		{"黑色+白色 XL", candidateSpecs([]string{"黑色", "白色"}, []string{"XL"})},
+		{"黑色 XL 多余描述", candidateSpecs([]string{"黑色"}, []string{"XL"})},
+		{"黑色 XL 建議55.5-60.0公斤穿", candidateSpecs([]string{"黑色"}, []string{"XL"})},
+		{"黑色，XL", candidateSpecs([]string{"黑色", "XL"}, []string{"XL", "黑色"})},
+		{"浅 灰色 XL", candidateSpecs([]string{"浅 灰色", "浅  灰色"}, []string{"XL"})},
+		{"白色,L", candidateSpecs([]string{"黑色"}, []string{"XL"})},
+		{"黑色 XL", `[{"name":"a","role":"color","values":[{"name":"黑色"}]},{"name":"b","role":"color","values":[{"name":"白色"}]},{"name":"s","role":"size","values":[{"name":"XL"}]}]`},
+		{"黑色 XL", `[{"name":"a","role":"other","values":[{"name":"黑色 XL"}]}]`},
+	} {
+		if got, want := sybimport.ParseWithCandidates(test.raw, test.specs), sybimport.Parse(test.raw); got != want {
+			t.Fatalf("raw %q changed: %+v vs %+v", test.raw, got, want)
+		}
+	}
+}
+
+func TestImportAndReparseReuseExistingCandidatesWithoutChangingMappings(t *testing.T) {
+	db := openTestDB(t)
+	ctx := context.Background()
+	detail := sybimport.DetailInput{ID: 25101, ProductID: 25101, ProductQty: 1, ProductPrice: 1, ProductSpec: "黑色+白色，XL", Raw: json.RawMessage(`{"productSpec":"黑色+白色，XL"}`)}
+	order := sybimport.OrderInput{Code: "test-251", ShopName: "test"}
+	first, err := sybimport.ApplyDetail(ctx, db, order, detail)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if first.SYBProduct.ParseStatus != models.SYBParseStatusUncertain {
+		t.Fatal("must not self-confirm without candidates")
+	}
+	specs := candidateSpecs([]string{"黑色+白色"}, []string{"XL"})
+	if err := db.Model(&models.ShopeeProduct{}).Where("id = ?", *first.SYBProduct.ShopeeProductID).Update("specs_json", specs).Error; err != nil {
+		t.Fatal(err)
+	}
+	outcome, err := sybimport.Reparse(ctx, db, first.SYBProduct.ID, false)
+	if err != nil || outcome.NewStatus != models.SYBParseStatusSuccess {
+		t.Fatalf("reparse: %+v %v", outcome, err)
+	}
+	second, err := sybimport.ApplyDetail(ctx, db, order, detail)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if second.SYBProduct.TargetColor != "黑色+白色" || second.SYBProduct.TargetSize != "XL" || second.SYBProduct.ParseStatus != models.SYBParseStatusSuccess {
+		t.Fatalf("import differs: %+v", second.SYBProduct)
+	}
+	if second.SYBProduct.RawJSON != first.SYBProduct.RawJSON {
+		t.Fatal("raw source changed")
+	}
+	if second.ShopeeProduct.SpecsJSON != specs {
+		t.Fatal("candidate confirmation must not rewrite archive dimensions or mappings")
+	}
+	if _, err := sybimport.ManualCorrect(ctx, db, first.SYBProduct.ID, "人工颜色", "人工尺码"); err != nil {
+		t.Fatal(err)
+	}
+	outcome, err = sybimport.Reparse(ctx, db, first.SYBProduct.ID, false)
+	if err != nil || outcome.Outcome != sybimport.ReparseOutcomeSkippedManual {
+		t.Fatalf("manual overwritten: %+v %v", outcome, err)
+	}
+	third, err := sybimport.ApplyDetail(ctx, db, order, detail)
+	if err != nil || third.SYBProduct.TargetColor != "人工颜色" {
+		t.Fatalf("reimport overwrote manual: %v", err)
+	}
+}

server/app/goauto/sybimport/reparse.go

@@ -77,6 +77,14 @@ func Reparse(ctx context.Context, db *gorm.DB, sybProductID uint64, force bool)
 			return fmt.Errorf("syb product %d: stored raw json is not decodable: %w", record.ID, err)
 		}
 		parsed := Parse(raw.ProductSpec)
+		if parsed.Status == models.SYBParseStatusUncertain && record.ShopeeProductID != nil {
+			var product models.ShopeeProduct
+			if err := tx.First(&product, *record.ShopeeProductID).Error; err == nil {
+				parsed = ParseWithCandidates(raw.ProductSpec, product.SpecsJSON)
+			} else if !errors.Is(err, gorm.ErrRecordNotFound) {
+				return err
+			}
+		}
 		outcome.NewStatus = parsed.Status
 
 		if parsed.Color == record.TargetColor && parsed.Size == record.TargetSize && parsed.Status == record.ParseStatus {