sm-seo-plugin/classes/field_types.go

package classes

import (
	"context"
	"errors"
	"fmt"
	"net/url"
	"strings"

	"git.golem15.com/golem15/sm-seo-plugin/classes/schema"
	"git.golem15.com/golem15/sm-seo-plugin/models"
	"git.golem15.com/golem15/summercms/modules/backpack"
	"git.golem15.com/golem15/summercms/modules/pact"
)

// The elements of the SEO field types and widgets (assets/src/elements).
const (
	ElementText         = "golem15-seo-text"
	ElementRobots       = "golem15-seo-robots"
	ElementSERP         = "golem15-seo-serp"
	ElementImage        = "golem15-seo-image"
	ElementRobotsTxt    = "golem15-seo-robotstxt"
	ElementRobotsTester = "golem15-seo-robotstester"
	ElementSchema       = "golem15-seo-schema"
	ElementGraph        = "golem15-seo-graph"
	ElementParent       = "golem15-seo-parent"
)

// The element bundle every SEO field type loads.
var (
	elementScripts = []string{"assets/dist/seo.js"}
	elementStyles  = []string{"assets/dist/seo.css"}
)

// errRobotsInvalid is the refusal of a malformed robots value.
const errRobotsInvalid = "golem15.seo::lang.robots.invalid"

// errRobotsTooLarge refuses a robots.txt over MaxRobotsBytes.
const errRobotsTooLarge = "golem15.seo::lang.robots.too_large"

// FieldTypes returns the SEO admin field types:
//   - seotext and mlseotext (E1): text with a live length counter; option
//     measure: title (input) or description (textarea);
//   - seorobots and mlseorobots (E2): the robots directives editor; option
//     scope: record, rule or type; Decode refuses a malformed value;
//   - seoimage (E7): an uploaded social image stored in the media bucket,
//     its URL as the value (D-100); uploads need permission;
//   - robotstxt (E4): the robots.txt editor with live lint; Decode refuses
//     a file over 500 KB (D-22).
//
// app is asked for the booted application at request time.
func FieldTypes(app func() *backpack.App, permission string) []pact.FormFieldType {
	text := func(name string, translatable bool) pact.FormFieldType {
		return pact.FormFieldType{
			Name: name, Element: ElementText, Translatable: translatable,
			Scripts: elementScripts, Styles: elementStyles,
			Options: []string{"measure"}, CheckOptions: checkMeasure, Decode: decodeText,
		}
	}
	robots := func(name string, translatable bool) pact.FormFieldType {
		return pact.FormFieldType{
			Name: name, Element: ElementRobots, Translatable: translatable,
			Scripts: elementScripts, Styles: elementStyles,
			Options: []string{"scope"}, CheckOptions: checkScope, Decode: decodeRobots,
		}
	}
	return []pact.FormFieldType{
		text("seotext", false),
		text("mlseotext", true),
		robots("seorobots", false),
		robots("mlseorobots", true),
		{
			Name: "seoimage", Element: ElementImage,
			Scripts: elementScripts, Styles: elementStyles,
			Decode: decodeImage, Upload: UploadImage(app, permission),
		},
		{
			Name: "robotstxt", Element: ElementRobotsTxt,
			Scripts: elementScripts, Styles: elementStyles,
			Decode: decodeRobotsTxt,
		},
		{
			Name: "seoschema", Element: ElementSchema,
			Scripts: elementScripts, Styles: elementStyles,
			Options: []string{"kind", "contentType"}, CheckOptions: checkKind,
			Decode: decodeSchema(app),
		},
		{
			Name: "seoparent", Element: ElementParent,
			Scripts: elementScripts, Styles: elementStyles,
			Options: []string{"contentType"},
			Decode:  decodeParent(app),
		},
	}
}

// checkKind accepts the content kinds of pact.ContentDescriptor.
func checkKind(options map[string]any) error {
	if raw, ok := options["kind"]; ok {
		switch s, _ := raw.(string); s {
		case "", schema.KindArticle, schema.KindCollection, schema.KindProfile, schema.KindPage:
		default:
			return fmt.Errorf("kind must be article, collection, profile or page, got %v", raw)
		}
	}
	return nil
}

// optionString reads a string option.
func optionString(options map[string]any, key string) string {
	s, _ := options[key].(string)
	return s
}

// SchemaRefusal renders a refused seo_schema value with the schema.* copy
// in the request locale (UI-SPEC C17).
func (s *Service) SchemaRefusal(ctx context.Context, kind string, err error) string {
	var se *schema.Error
	if !errors.As(err, &se) {
		return s.PhraseCtx(ctx, "golem15.seo::lang.schema.invalid_json", map[string]string{"error": err.Error()})
	}
	switch se.Code {
	case schema.CodeInvalidJSON:
		return s.PhraseCtx(ctx, "golem15.seo::lang.schema.invalid_json", map[string]string{"error": se.Detail})
	case schema.CodeTypeUnknown:
		return s.PhraseCtx(ctx, "golem15.seo::lang.schema.type_unknown", map[string]string{"type": kind})
	case schema.CodeFieldInvalid:
		label, message := "golem15.seo::lang.schema.field_price", "golem15.seo::lang.schema.price_invalid"
		if se.Field == "currency" {
			label, message = "golem15.seo::lang.schema.field_currency", "golem15.seo::lang.schema.currency_invalid"
		}
		return s.PhraseCtx(ctx, "golem15.seo::lang.schema.field_invalid", map[string]string{
			"field":   s.PhraseCtx(ctx, label, nil),
			"message": s.PhraseCtx(ctx, message, nil),
		})
	}
	keys := map[string]string{
		schema.CodeNotObject: "golem15.seo::lang.schema.not_object",
		schema.CodeTooLarge:  "golem15.seo::lang.schema.too_large",
		schema.CodeTooDeep:   "golem15.seo::lang.schema.too_deep",
		schema.CodeScriptTag: "golem15.seo::lang.schema.script_tag",
	}
	return s.PhraseCtx(ctx, keys[se.Code], nil)
}

// SchemaVariant is the registry kind of a record of contentType, "" for
// other types.
func SchemaVariant(record any) string {
	if p, ok := record.(*models.Page); ok && p != nil {
		return p.Kind
	}
	return ""
}

// decodeSchema checks and normalises the seoschema value (E8): a JSON
// string or "" (UI-SPEC C17); every refusal answers the schema.* copy.
func decodeSchema(app func() *backpack.App) func(context.Context, pact.FormFieldValue) (pact.FormFieldDecoded, error) {
	return func(ctx context.Context, in pact.FormFieldValue) (pact.FormFieldDecoded, error) {
		raw, ok := stringValue(in.Value)
		if !ok {
			return pact.FormFieldDecoded{}, &pact.FormFieldError{Messages: []string{"golem15.seo::lang.schema.not_object"}}
		}
		kind := optionString(in.Options, "kind")
		stored, _, err := schema.Normalize(kind, SchemaVariant(in.Record), raw)
		if err != nil {
			svc, serr := NewService(app())
			if serr != nil {
				return pact.FormFieldDecoded{}, &pact.FormFieldError{Messages: []string{err.Error()}}
			}
			return pact.FormFieldDecoded{}, &pact.FormFieldError{Messages: []string{svc.SchemaRefusal(ctx, kind, err)}}
		}
		return pact.FormFieldDecoded{Value: stored}, nil
	}
}

// decodeParent accepts "" (the first parent) or one of the record's
// parents as <type>:<id>.
func decodeParent(app func() *backpack.App) func(context.Context, pact.FormFieldValue) (pact.FormFieldDecoded, error) {
	return func(ctx context.Context, in pact.FormFieldValue) (pact.FormFieldDecoded, error) {
		value := in.Value
		if m, isMap := value.(map[string]any); isMap {
			// The loaded {selected, options} sent back unchanged.
			value = m["selected"]
		}
		raw, ok := stringValue(value)
		raw = strings.TrimSpace(raw)
		if !ok {
			return pact.FormFieldDecoded{}, &pact.FormFieldError{Messages: []string{"golem15.seo::lang.api.value_invalid"}}
		}
		if raw == "" {
			return pact.FormFieldDecoded{Value: ""}, nil
		}
		svc, err := NewService(app())
		if err != nil {
			return pact.FormFieldDecoded{}, err
		}
		id := ""
		if in.Record != nil && in.DB != nil {
			id, _ = RecordKey(in.DB, in.Record)
		}
		for _, opt := range svc.ParentOptions(ctx, optionString(in.Options, "contentType"), id, "") {
			if opt.Value == raw {
				return pact.FormFieldDecoded{Value: raw}, nil
			}
		}
		return pact.FormFieldDecoded{}, &pact.FormFieldError{Messages: []string{"golem15.seo::lang.api.value_invalid"}}
	}
}

// decodeRobotsTxt stores a robots.txt with Unix line ends and refuses one
// over MaxRobotsBytes (robots.too_large); lint findings are warnings in
// the element and never block a save.
func decodeRobotsTxt(_ context.Context, in pact.FormFieldValue) (pact.FormFieldDecoded, error) {
	s, ok := stringValue(in.Value)
	if !ok {
		return pact.FormFieldDecoded{}, &pact.FormFieldError{Messages: []string{"The " + in.Field + " field must be text."}}
	}
	s = strings.ReplaceAll(s, "\r\n", "\n")
	if len(s) > MaxRobotsBytes {
		return pact.FormFieldDecoded{}, &pact.FormFieldError{Messages: []string{errRobotsTooLarge}}
	}
	return pact.FormFieldDecoded{Value: s}, nil
}

func checkMeasure(options map[string]any) error {
	if raw, ok := options["measure"]; ok {
		if s, _ := raw.(string); s != "title" && s != "description" {
			return fmt.Errorf("measure must be title or description, got %v", raw)
		}
	}
	return nil
}

func checkScope(options map[string]any) error {
	if raw, ok := options["scope"]; ok {
		if s, _ := raw.(string); s != "record" && s != "rule" && s != "type" {
			return fmt.Errorf("scope must be record, rule or type, got %v", raw)
		}
	}
	return nil
}

// stringValue reads a submitted value as text; nil is "".
func stringValue(v any) (string, bool) {
	switch s := v.(type) {
	case nil:
		return "", true
	case string:
		return s, true
	}
	return "", false
}

func decodeText(_ context.Context, in pact.FormFieldValue) (pact.FormFieldDecoded, error) {
	s, ok := stringValue(in.Value)
	if !ok {
		return pact.FormFieldDecoded{}, &pact.FormFieldError{Messages: []string{"The " + in.Field + " field must be text."}}
	}
	return pact.FormFieldDecoded{Value: strings.TrimSpace(s)}, nil
}

// decodeRobots stores a robots value in the fixed order; a malformed value
// (reachable only through MCP or an import) is refused.
func decodeRobots(_ context.Context, in pact.FormFieldValue) (pact.FormFieldDecoded, error) {
	s, ok := stringValue(in.Value)
	if !ok {
		return pact.FormFieldDecoded{}, &pact.FormFieldError{Messages: []string{errRobotsInvalid}}
	}
	normal, err := NormalizeRobots(s)
	if err != nil {
		return pact.FormFieldDecoded{}, &pact.FormFieldError{Messages: []string{errRobotsInvalid}}
	}
	return pact.FormFieldDecoded{Value: normal}, nil
}

// decodeImage accepts an empty value, a path or an http(s) URL.
func decodeImage(_ context.Context, in pact.FormFieldValue) (pact.FormFieldDecoded, error) {
	s, ok := stringValue(in.Value)
	s = strings.TrimSpace(s)
	if !ok || !ValidImageURL(s) {
		return pact.FormFieldDecoded{}, &pact.FormFieldError{Messages: []string{errImageURL}}
	}
	return pact.FormFieldDecoded{Value: s}, nil
}

// ValidURLRef reports whether v is a path ("/x", not "//x" or "/\x",
// which browsers read as another host) or an absolute http(s) URL with a
// host, without spaces or control characters.
func ValidURLRef(v string) bool {
	if v == "" || strings.ContainsFunc(v, func(r rune) bool { return r <= ' ' || r == 0x7f }) {
		return false
	}
	if strings.HasPrefix(v, "/") {
		return !strings.HasPrefix(v, "//") && !strings.HasPrefix(v, `/\`)
	}
	u, err := url.Parse(v)
	return err == nil && (u.Scheme == "http" || u.Scheme == "https") && u.Host != ""
}

// SchemaProblem is a refused seo_schema value as an API problem.
func SchemaProblem(kind string, err error) models.Problem {
	var se *schema.Error
	if !errors.As(err, &se) {
		return models.Problem{Key: "golem15.seo::lang.schema.invalid_json", Params: map[string]string{"error": err.Error()}}
	}
	switch se.Code {
	case schema.CodeInvalidJSON:
		return models.Problem{Key: "golem15.seo::lang.schema.invalid_json", Params: map[string]string{"error": se.Detail}}
	case schema.CodeTypeUnknown:
		return models.Problem{Key: "golem15.seo::lang.schema.type_unknown", Params: map[string]string{"type": kind}}
	case schema.CodeFieldInvalid:
		if se.Field == "currency" {
			return models.Problem{Key: "golem15.seo::lang.schema.currency_invalid"}
		}
		return models.Problem{Key: "golem15.seo::lang.schema.price_invalid"}
	case schema.CodeNotObject:
		return models.Problem{Key: "golem15.seo::lang.schema.not_object"}
	case schema.CodeTooLarge:
		return models.Problem{Key: "golem15.seo::lang.schema.too_large"}
	case schema.CodeTooDeep:
		return models.Problem{Key: "golem15.seo::lang.schema.too_deep"}
	}
	return models.Problem{Key: "golem15.seo::lang.schema.script_tag"}
}