feat: add multi-source breed encyclopedia and two-hour refresh sessions

pets-be/internal/breeds/breeds.go

@@ -0,0 +1,102 @@
+package breeds
+
+import (
+	_ "embed"
+	"encoding/json"
+	"fmt"
+	"strings"
+	"unicode"
+
+	"github.com/mozillazg/go-pinyin"
+	"github.com/sundynix/pets-be/internal/model"
+	"gorm.io/gorm"
+	"gorm.io/gorm/clause"
+)
+
+//go:embed data.json
+var data []byte
+
+// SortFields provides both full pinyin and initials, retaining Latin breed names.
+func SortFields(name string) (string, string, string) {
+	args := pinyin.NewArgs()
+	args.Fallback = func(r rune, _ pinyin.Args) []string {
+		if unicode.IsLetter(r) || unicode.IsDigit(r) {
+			return []string{strings.ToLower(string(r))}
+		}
+		return nil
+	}
+	words := pinyin.LazyPinyin(name, args)
+	// Breed-name phrases with common alternative pronunciations.
+	if strings.HasPrefix(name, "重庆") && len(words) > 0 {
+		words[0] = "chong"
+	}
+	if strings.HasPrefix(name, "长毛") && len(words) > 0 {
+		words[0] = "chang"
+	}
+	full := strings.Join(words, "")
+	var initials strings.Builder
+	for _, word := range words {
+		if word != "" {
+			initials.WriteByte(word[0])
+		}
+	}
+	initial := "#"
+	if len(full) > 0 && full[0] >= 'a' && full[0] <= 'z' {
+		initial = strings.ToUpper(full[:1])
+	}
+	return full, initials.String(), initial
+}
+
+func Defaults() ([]model.Breed, error) {
+	var rows []struct {
+		Origin                string `json:"origin"`
+		BreedGroup            string `json:"breed_group"`
+		FCINumber             string `json:"fci_number"`
+		EnglishName           string `json:"english_name"`
+		Height                string `json:"height"`
+		Weight                string `json:"weight"`
+		Summary               string `json:"summary"`
+		Temperament           string `json:"temperament"`
+		Traits                string `json:"traits"`
+		CareNotes             string `json:"care_notes"`
+		EncyclopediaUpdatedAt string `json:"encyclopedia_updated_at"`
+		Name                  string `json:"name"`
+		Aliases               string `json:"aliases"`
+		Source                string `json:"source"`
+		SourceKey             string `json:"source_key"`
+		SourceURL             string `json:"source_url"`
+	}
+	if err := json.Unmarshal(data, &rows); err != nil {
+		return nil, err
+	}
+	result := make([]model.Breed, 0, len(rows))
+	for _, row := range rows {
+		full, short, initial := SortFields(row.Name)
+		key := row.SourceKey
+		result = append(result, model.Breed{Origin: row.Origin, BreedGroup: row.BreedGroup, FCINumber: row.FCINumber, EnglishName: row.EnglishName, Height: row.Height, Weight: row.Weight, Summary: row.Summary, Temperament: row.Temperament, Traits: row.Traits, CareNotes: row.CareNotes, EncyclopediaUpdatedAt: row.EncyclopediaUpdatedAt, Name: row.Name, Aliases: row.Aliases, Pinyin: full, PinyinInitials: short, Initial: initial, Enabled: true, Source: row.Source, SourceKey: &key, SourceURL: row.SourceURL})
+	}
+	return result, nil
+}
+
+// Existing entries (including edits and disabled breeds) are never overwritten.
+func Seed(db *gorm.DB) (int64, error) {
+	rows, err := Defaults()
+	if err != nil {
+		return 0, fmt.Errorf("load breeds: %w", err)
+	}
+	result := db.Clauses(clause.OnConflict{DoNothing: true}).CreateInBatches(rows, 100)
+	if result.Error != nil {
+		return result.RowsAffected, result.Error
+	}
+	// Backfill newly introduced encyclopedia fields exactly once on untouched seed rows.
+	// Future starts never overwrite administrator changes.
+	for _, row := range rows {
+		if row.Summary == "" {
+			continue
+		}
+		if err := db.Model(&model.Breed{}).Where("source_key = ? AND name = ? AND (encyclopedia_updated_at IS NULL OR encyclopedia_updated_at = '') AND updated_at = created_at", row.SourceKey, row.Name).Updates(map[string]any{"origin": row.Origin, "breed_group": row.BreedGroup, "fci_number": row.FCINumber, "english_name": row.EnglishName, "height": row.Height, "weight": row.Weight, "summary": row.Summary, "temperament": row.Temperament, "traits": row.Traits, "care_notes": row.CareNotes, "encyclopedia_updated_at": row.EncyclopediaUpdatedAt}).Error; err != nil {
+			return result.RowsAffected, err
+		}
+	}
+	return result.RowsAffected, enrich(db)
+}