sm-seo-plugin/classes/robots_write.go

package classes

import (
	"context"
	"errors"
	"strconv"
	"strings"
	"time"

	"git.golem15.com/golem15/sm-seo-plugin/models"
	"gorm.io/gorm"
	"gorm.io/gorm/clause"
)

// RobotsRow returns robots row 1 inside db, creating it when the
// migration's row is gone (first or create, safe against a concurrent
// first request).
func RobotsRow(db *gorm.DB) (models.Robots, error) {
	var rows []models.Robots
	if err := db.Where("id = ?", 1).Limit(1).Find(&rows).Error; err != nil {
		return models.Robots{}, err
	}
	if len(rows) > 0 {
		return rows[0], nil
	}
	row := models.Robots{ID: 1, Bots: models.BotSwitches{}}
	if err := db.Clauses(clause.OnConflict{DoNothing: true}).Create(&row).Error; err != nil {
		return models.Robots{}, err
	}
	if err := db.Where("id = ?", 1).Limit(1).Find(&rows).Error; err != nil {
		return models.Robots{}, err
	}
	if len(rows) == 0 {
		return models.Robots{}, errors.New("golem15.seo: robots row is missing")
	}
	return rows[0], nil
}

// RobotsValues are the robots row's values for history: the editor's text
// and every listed bot's switch.
func RobotsValues(r models.Robots) map[string]any {
	bots := map[string]any{}
	for _, b := range Bots {
		bots[b.Token] = r.Bots.Allowed(b.Token)
	}
	return map[string]any{"content": r.Content, "bots": bots, "llms_enabled": r.LlmsEnabled, "llms_intro": r.LlmsIntro}
}

// RobotsWrite is one change of robots.txt: the editor's text when Content
// is set, and the switches of the bots listed in Bots (by token).
type RobotsWrite struct {
	Content     *string
	Bots        map[string]bool
	LlmsEnabled *bool
	LlmsIntro   *string
}

// SaveRobots writes robots.txt inside tx and records it in history
// (D-22, D-25). A text over MaxRobotsBytes and a token that is not a
// listed AI bot are refused with a *ProblemsError. It returns the stored
// row and the noindex addresses Googlebot may no longer fetch under it
// (the robots.blocks_noindex warning).
func (s *Service) SaveRobots(ctx context.Context, tx *gorm.DB, in RobotsWrite) (models.Robots, []string, error) {
	row, err := RobotsRow(tx)
	if err != nil {
		return models.Robots{}, nil, err
	}
	problems := map[string][]models.Problem{}
	if in.Content != nil && len(*in.Content) > MaxRobotsBytes {
		problems["content"] = append(problems["content"], models.Problem{Key: "golem15.seo::lang.robots.too_large"})
	}
	for token := range in.Bots {
		if _, ok := BotByToken(token); !ok {
			problems["bots."+token] = append(problems["bots."+token], models.Problem{Key: "golem15.seo::lang.api.bot_unknown"})
		}
	}
	if len(problems) > 0 {
		return models.Robots{}, nil, &ProblemsError{Problems: problems}
	}
	before := RobotsValues(row)
	if in.Content != nil {
		row.Content = *in.Content
	}
	bots := models.BotSwitches{}
	for token, on := range row.Bots {
		bots[token] = on
	}
	for token, on := range in.Bots {
		b, _ := BotByToken(token)
		bots[b.Token] = on
	}
	row.Bots = bots
	if in.LlmsEnabled != nil {
		row.LlmsEnabled = *in.LlmsEnabled
	}
	if in.LlmsIntro != nil {
		row.LlmsIntro = strings.TrimSpace(*in.LlmsIntro)
	}
	now := time.Now().UTC()
	row.UpdatedAt = &now
	if err := tx.Model(&models.Robots{}).Where("id = ?", row.ID).Updates(map[string]any{
		"content": row.Content, "bots": row.Bots, "llms_enabled": row.LlmsEnabled, "llms_intro": row.LlmsIntro, "updated_at": now,
	}).Error; err != nil {
		return models.Robots{}, nil, err
	}
	if err := s.Record(ctx, tx, Change{
		Kind:      models.SubjectRobots,
		SubjectID: strconv.FormatUint(uint64(row.ID), 10),
		What:      s.Phrase("golem15.seo::lang.history.what_robots", nil),
		Before:    before,
		After:     RobotsValues(row),
	}); err != nil {
		return models.Robots{}, nil, err
	}
	blocked, err := s.BlockedNoindex(tx, row.Content, row.Bots)
	if err != nil {
		return models.Robots{}, nil, err
	}
	return row, blocked, nil
}