descrybe/apps/api/internal/processing/legacy_convert.go

package processing

import (
	"fmt"
	"strings"
)

// isDescriptionFieldKey reports keys whose values should be plain text scalars
// (not JSON arrays / nested HTML wrappers) for process + V1 poll contracts.
func isDescriptionFieldKey(key string) bool {
	switch strings.ToLower(strings.TrimSpace(key)) {
	case "description", "desc", "body", "short_description", "shortdescription",
		"processed_description", "long_description", "longdescription":
		return true
	default:
		return false
	}
}

// PlainDescriptionFromAny coerces legacy description shapes to a single plain
// string: string, {#text}, or arrays of those (joined with newlines). HTML is
// stripped via v1PlainDescription — use for meta/SEO and feed normalize only.
// V1 process poll / formula bodies use DescriptionFromAny (keeps markup).
func PlainDescriptionFromAny(v any) string {
	if v == nil {
		return ""
	}
	switch t := v.(type) {
	case string:
		return v1PlainDescription(t)
	case []string:
		parts := make([]string, 0, len(t))
		for _, s := range t {
			if p := v1PlainDescription(s); p != "" {
				parts = append(parts, p)
			}
		}
		return strings.Join(parts, "\n")
	case []any:
		parts := make([]string, 0, len(t))
		for _, item := range t {
			if p := PlainDescriptionFromAny(item); p != "" {
				parts = append(parts, p)
			}
		}
		return strings.Join(parts, "\n")
	case map[string]any:
		for _, k := range []string{"#text", "text", "value", "description", "body"} {
			if p := PlainDescriptionFromAny(t[k]); p != "" {
				return p
			}
		}
		return ""
	default:
		s := strings.TrimSpace(fmt.Sprint(t))
		if s == "" || s == "<nil>" {
			return ""
		}
		return v1PlainDescription(s)
	}
}

// DescriptionFromAny coerces legacy description shapes to a single string while
// preserving formula HTML (h1/h2/p/ul/…). Unwraps JSON arrays / #text wrappers
// but does not strip tags — used by V1 process item projection and catalog
// normalize when enhance/formula HTML must reach clients and stay in DB.
func DescriptionFromAny(v any) string {
	if v == nil {
		return ""
	}
	switch t := v.(type) {
	case string:
		return v1PreserveDescription(t)
	case []string:
		parts := make([]string, 0, len(t))
		for _, s := range t {
			if p := v1PreserveDescription(s); p != "" {
				parts = append(parts, p)
			}
		}
		return strings.Join(parts, "\n")
	case []any:
		parts := make([]string, 0, len(t))
		for _, item := range t {
			if p := DescriptionFromAny(item); p != "" {
				parts = append(parts, p)
			}
		}
		return strings.Join(parts, "\n")
	case map[string]any:
		for _, k := range []string{"#text", "text", "value", "description", "body"} {
			if p := DescriptionFromAny(t[k]); p != "" {
				return p
			}
		}
		return ""
	default:
		s := strings.TrimSpace(fmt.Sprint(t))
		if s == "" || s == "<nil>" {
			return ""
		}
		return v1PreserveDescription(s)
	}
}

// coerceScalarText flattens arrays / #text wrappers into a trimmed string without
// HTML stripping (names, brands, stock labels, …).
func coerceScalarText(v any) string {
	if v == nil {
		return ""
	}
	switch t := v.(type) {
	case string:
		return strings.TrimSpace(t)
	case []string:
		parts := make([]string, 0, len(t))
		for _, s := range t {
			s = strings.TrimSpace(s)
			if s != "" {
				parts = append(parts, s)
			}
		}
		return strings.Join(parts, " ")
	case []any:
		parts := make([]string, 0, len(t))
		for _, item := range t {
			if s := coerceScalarText(item); s != "" {
				parts = append(parts, s)
			}
		}
		return strings.Join(parts, " ")
	case map[string]any:
		for _, k := range []string{"#text", "text", "value"} {
			if s := coerceScalarText(t[k]); s != "" {
				return s
			}
		}
		return ""
	case float64, float32, int, int64, int32, bool:
		s := strings.TrimSpace(fmt.Sprint(t))
		if s == "<nil>" {
			return ""
		}
		return s
	default:
		if n, ok := t.(jsonNumberStringer); ok {
			return strings.TrimSpace(n.String())
		}
		s := strings.TrimSpace(fmt.Sprint(t))
		if s == "" || s == "<nil>" {
			return ""
		}
		return s
	}
}