sm-bm-plugin/classes/seed.go

package classes

import (
	"bytes"
	"context"
	"encoding/json"
	"errors"
	"fmt"
	"io"
	"slices"
	"strings"
	"time"

	"git.golem15.com/golem15/summercms/modules/lagoon"
	"git.golem15.com/jakub/sm-bm-plugin/models"
	"gorm.io/gorm"
	"gorm.io/gorm/clause"
)

// SeedFile is the JSON document bm:seed reads. Every top-level array is
// optional. Courses carry a seed-local key that class sessions, deadlines
// and readings use to point at them; meetings are referenced by number.
type SeedFile struct {
	Meetings      []SeedMeeting  `json:"meetings"`
	Courses       []SeedCourse   `json:"courses"`
	ClassSessions []SeedSession  `json:"class_sessions"`
	Deadlines     []SeedDeadline `json:"deadlines"`
	Readings      []SeedReading  `json:"readings"`
}

// SeedMeeting is one meeting; its identity is Number.
type SeedMeeting struct {
	Number   int     `json:"number"`
	StartsOn string  `json:"starts_on"`
	EndsOn   string  `json:"ends_on"`
	Notes    *string `json:"notes"`
}

// SeedCourse is one course; its identity is Name, Form and Semester. Key is
// only a reference inside the file.
type SeedCourse struct {
	Key           string   `json:"key"`
	Code          *string  `json:"code"`
	Name          string   `json:"name"`
	Form          string   `json:"form"`
	Lecturer      *string  `json:"lecturer"`
	Semester      int      `json:"semester"`
	ECTS          *float64 `json:"ects"`
	Assessment    *string  `json:"assessment"`
	ElectiveGroup *string  `json:"elective_group"`
	Description   *string  `json:"description"`
	SyllabusURL   *string  `json:"syllabus_url"`
	SortOrder     int      `json:"sort_order"`
}

// SeedSession is one class session; its identity is the course and StartsAt.
// Course is a course key, Meeting an optional meeting number.
type SeedSession struct {
	Course   string  `json:"course"`
	Meeting  *int    `json:"meeting"`
	StartsAt string  `json:"starts_at"`
	EndsAt   string  `json:"ends_at"`
	Room     *string `json:"room"`
	Kind     *string `json:"kind"`
	IsOnline bool    `json:"is_online"`
	Notes    *string `json:"notes"`
}

// SeedDeadline is one deadline; its identity is Title and DueAt. Course is
// an optional course key.
type SeedDeadline struct {
	Title       string  `json:"title"`
	DueAt       string  `json:"due_at"`
	Course      *string `json:"course"`
	Kind        string  `json:"kind"`
	Description *string `json:"description"`
}

// SeedReading is one reading; its identity is the course and Title.
type SeedReading struct {
	Course     string  `json:"course"`
	Title      string  `json:"title"`
	Author     *string `json:"author"`
	Source     *string `json:"source"`
	URL        *string `json:"url"`
	IsRequired bool    `json:"is_required"`
	SortOrder  int     `json:"sort_order"`
}

// SeedCounts counts the rows of one kind a seed created, updated, left
// unchanged or (with SeedOptions.Prune) deleted.
type SeedCounts struct {
	Created   int
	Updated   int
	Unchanged int
	Deleted   int
}

// SeedOptions changes how a seed is applied.
type SeedOptions struct {
	// Prune deletes the courses of the file's semesters, and the class
	// sessions and readings of the file's courses, that the file does not
	// list. Meetings and deadlines are never pruned.
	Prune bool
	// DryRun applies the seed in a transaction that is rolled back, so the
	// summary shows what would change and nothing is written.
	DryRun bool
}

// errDryRun rolls back a dry-run transaction.
var errDryRun = errors.New("bm:seed: dry run")

func (c *SeedCounts) record(created, changed bool) {
	switch {
	case created:
		c.Created++
	case changed:
		c.Updated++
	default:
		c.Unchanged++
	}
}

// SeedSummary is the outcome of a seed, per kind.
type SeedSummary struct {
	Meetings      SeedCounts
	Courses       SeedCounts
	ClassSessions SeedCounts
	Deadlines     SeedCounts
	Readings      SeedCounts
	// Removed describes each row Prune deleted, for the command output.
	Removed []string
}

// Rows returns the summary as table rows: kind, created, updated,
// unchanged, deleted.
func (s SeedSummary) Rows() [][]string {
	row := func(name string, c SeedCounts) []string {
		return []string{name, fmt.Sprint(c.Created), fmt.Sprint(c.Updated), fmt.Sprint(c.Unchanged), fmt.Sprint(c.Deleted)}
	}
	return [][]string{
		row("meetings", s.Meetings),
		row("courses", s.Courses),
		row("class_sessions", s.ClassSessions),
		row("deadlines", s.Deadlines),
		row("readings", s.Readings),
	}
}

// ParseSeed decodes a seed document. Unknown fields are an error, so a typo
// in a key never drops data silently.
func ParseSeed(r io.Reader) (*SeedFile, error) {
	body, err := io.ReadAll(r)
	if err != nil {
		return nil, err
	}
	dec := json.NewDecoder(bytes.NewReader(body))
	dec.DisallowUnknownFields()
	var f SeedFile
	if err := dec.Decode(&f); err != nil {
		return nil, fmt.Errorf("bm:seed: %w", err)
	}
	if dec.More() {
		return nil, errors.New("bm:seed: trailing data after the JSON document")
	}
	return &f, nil
}

// Validate checks the whole document before anything is written: required
// fields, allowed values, dates and times, duplicate identities and that
// every course key and meeting number a record points at exists in the
// file. Each error names the record.
func (f *SeedFile) Validate() error {
	var errs []error
	add := func(format string, args ...any) { errs = append(errs, fmt.Errorf(format, args...)) }

	meetings := map[int]bool{}
	for i, m := range f.Meetings {
		at := fmt.Sprintf("meetings[%d] (number %d)", i, m.Number)
		if m.Number < 1 {
			add("%s: number must be a positive integer", at)
		}
		if meetings[m.Number] {
			add("%s: duplicate meeting number", at)
		}
		meetings[m.Number] = true
		start, err1 := models.ParseDate(m.StartsOn)
		end, err2 := models.ParseDate(m.EndsOn)
		if err1 != nil {
			add("%s: starts_on: %v", at, err1)
		}
		if err2 != nil {
			add("%s: ends_on: %v", at, err2)
		}
		if err1 == nil && err2 == nil && end.Time(nil).Before(start.Time(nil)) {
			add("%s: ends_on is before starts_on", at)
		}
	}

	keys := map[string]bool{}
	identities := map[string]bool{}
	for i, c := range f.Courses {
		at := fmt.Sprintf("courses[%d] (key %q)", i, c.Key)
		if strings.TrimSpace(c.Key) == "" {
			add("%s: key is required", at)
		} else if keys[c.Key] {
			add("%s: duplicate course key", at)
		}
		keys[c.Key] = true
		if strings.TrimSpace(c.Name) == "" {
			add("%s: name is required", at)
		}
		if !slices.Contains(models.CourseForms, c.Form) {
			add("%s: form %q is not one of %s", at, c.Form, strings.Join(models.CourseForms, ", "))
		}
		if c.Semester < 1 || c.Semester > 4 {
			add("%s: semester must be between 1 and 4", at)
		}
		id := fmt.Sprintf("%s\x00%s\x00%d", c.Name, c.Form, c.Semester)
		if identities[id] {
			add("%s: another course has the same name, form and semester", at)
		}
		identities[id] = true
	}

	sessions := map[string]bool{}
	for i, s := range f.ClassSessions {
		at := fmt.Sprintf("class_sessions[%d] (course %q, starts_at %s)", i, s.Course, s.StartsAt)
		if !keys[s.Course] {
			add("%s: unknown course key %q", at, s.Course)
		}
		if s.Meeting != nil && !meetings[*s.Meeting] {
			add("%s: unknown meeting number %d", at, *s.Meeting)
		}
		start, err1 := models.ParseDateTime(s.StartsAt)
		end, err2 := models.ParseDateTime(s.EndsAt)
		if err1 != nil {
			add("%s: starts_at: %v", at, err1)
		}
		if err2 != nil {
			add("%s: ends_at: %v", at, err2)
		}
		if err1 == nil && err2 == nil {
			if !end.After(start) {
				add("%s: ends_at is not after starts_at", at)
			}
			id := s.Course + "\x00" + start.UTC().String()
			if sessions[id] {
				add("%s: duplicate session of the course at this time", at)
			}
			sessions[id] = true
		}
	}

	deadlines := map[string]bool{}
	for i, d := range f.Deadlines {
		at := fmt.Sprintf("deadlines[%d] (%q)", i, d.Title)
		if strings.TrimSpace(d.Title) == "" {
			add("%s: title is required", at)
		}
		if d.Course != nil && !keys[*d.Course] {
			add("%s: unknown course key %q", at, *d.Course)
		}
		if !slices.Contains(models.DeadlineKinds, d.Kind) {
			add("%s: kind %q is not one of %s", at, d.Kind, strings.Join(models.DeadlineKinds, ", "))
		}
		due, err := models.ParseDateTime(d.DueAt)
		if err != nil {
			add("%s: due_at: %v", at, err)
			continue
		}
		id := d.Title + "\x00" + due.UTC().String()
		if deadlines[id] {
			add("%s: duplicate deadline (same title and due_at)", at)
		}
		deadlines[id] = true
	}

	readings := map[string]bool{}
	for i, r := range f.Readings {
		at := fmt.Sprintf("readings[%d] (course %q, %q)", i, r.Course, r.Title)
		if !keys[r.Course] {
			add("%s: unknown course key %q", at, r.Course)
		}
		if strings.TrimSpace(r.Title) == "" {
			add("%s: title is required", at)
		}
		id := r.Course + "\x00" + r.Title
		if readings[id] {
			add("%s: duplicate reading of the course", at)
		}
		readings[id] = true
	}
	return errors.Join(errs...)
}

// Seed validates f and upserts it in one transaction: meetings by number,
// courses by name, form and semester, class sessions by course and start
// time, deadlines by title and due time, readings by course and title. Rows
// absent from the file are left alone. On any error nothing is written.
func Seed(ctx context.Context, db *gorm.DB, f *SeedFile) (SeedSummary, error) {
	return SeedWith(ctx, db, f, SeedOptions{})
}

// SeedWith is Seed with options: pruning rows the file no longer lists and
// dry runs.
func SeedWith(ctx context.Context, db *gorm.DB, f *SeedFile, opts SeedOptions) (SeedSummary, error) {
	var sum SeedSummary
	if err := f.Validate(); err != nil {
		return sum, err
	}
	err := lagoon.Transaction(ctx, db, func(ctx context.Context, tx *gorm.DB) error {
		meetingIDs := map[int]uint{}
		for _, in := range f.Meetings {
			m := models.Meeting{}
			created, changed, err := upsert(tx, &m, clause.Eq{Column: "number", Value: in.Number}, func(m *models.Meeting) {
				m.Number = in.Number
				m.StartsOn, _ = models.ParseDate(in.StartsOn)
				m.EndsOn, _ = models.ParseDate(in.EndsOn)
				m.Notes = nz(in.Notes)
			})
			if err != nil {
				return fmt.Errorf("meeting %d: %w", in.Number, err)
			}
			sum.Meetings.record(created, changed)
			meetingIDs[in.Number] = m.ID
		}

		courseIDs := map[string]uint{}
		semesters := map[int]bool{}
		for _, in := range f.Courses {
			semesters[in.Semester] = true
			c := models.Course{}
			where := clause.And(
				clause.Eq{Column: "name", Value: in.Name},
				clause.Eq{Column: "form", Value: in.Form},
				clause.Eq{Column: "semester", Value: in.Semester},
			)
			created, changed, err := upsert(tx, &c, where, func(c *models.Course) {
				c.Code = nz(in.Code)
				c.Name = in.Name
				c.Form = in.Form
				c.Lecturer = nz(in.Lecturer)
				c.Semester = in.Semester
				c.ECTS = in.ECTS
				c.Assessment = nz(in.Assessment)
				c.ElectiveGroup = nz(in.ElectiveGroup)
				c.Description = nz(in.Description)
				c.SyllabusURL = nz(in.SyllabusURL)
				c.SortOrder = in.SortOrder
			})
			if err != nil {
				return fmt.Errorf("course %q: %w", in.Key, err)
			}
			sum.Courses.record(created, changed)
			courseIDs[in.Key] = c.ID
		}

		keptSessions := []uint{0}
		for _, in := range f.ClassSessions {
			start, _ := models.ParseDateTime(in.StartsAt)
			s := models.ClassSession{}
			where := clause.And(
				clause.Eq{Column: "course_id", Value: courseIDs[in.Course]},
				clause.Eq{Column: "starts_at", Value: start.UTC()},
			)
			created, changed, err := upsert(tx, &s, where, func(s *models.ClassSession) {
				courseID := courseIDs[in.Course]
				s.CourseID = &courseID
				s.MeetingID = nil
				if in.Meeting != nil {
					id := meetingIDs[*in.Meeting]
					s.MeetingID = &id
				}
				end, _ := models.ParseDateTime(in.EndsAt)
				setInstant(&s.StartsAt, start)
				setInstant(&s.EndsAt, end)
				s.Room = nz(in.Room)
				s.Kind = nz(in.Kind)
				s.IsOnline = in.IsOnline
				s.Notes = nz(in.Notes)
			})
			if err != nil {
				return fmt.Errorf("class session %s %s: %w", in.Course, in.StartsAt, err)
			}
			sum.ClassSessions.record(created, changed)
			keptSessions = append(keptSessions, s.ID)
		}

		for _, in := range f.Deadlines {
			due, _ := models.ParseDateTime(in.DueAt)
			d := models.Deadline{}
			where := clause.And(
				clause.Eq{Column: "title", Value: in.Title},
				clause.Eq{Column: "due_at", Value: due.UTC()},
			)
			created, changed, err := upsert(tx, &d, where, func(d *models.Deadline) {
				d.Title = in.Title
				setInstant(&d.DueAt, due)
				d.CourseID = nil
				if in.Course != nil {
					id := courseIDs[*in.Course]
					d.CourseID = &id
				}
				d.Kind = in.Kind
				d.Description = nz(in.Description)
			})
			if err != nil {
				return fmt.Errorf("deadline %q: %w", in.Title, err)
			}
			sum.Deadlines.record(created, changed)
		}

		keptReadings := []uint{0}
		for _, in := range f.Readings {
			r := models.Reading{}
			where := clause.And(
				clause.Eq{Column: "course_id", Value: courseIDs[in.Course]},
				clause.Eq{Column: "title", Value: in.Title},
			)
			created, changed, err := upsert(tx, &r, where, func(r *models.Reading) {
				courseID := courseIDs[in.Course]
				r.CourseID = &courseID
				r.Title = in.Title
				r.Author = nz(in.Author)
				r.Source = nz(in.Source)
				r.URL = nz(in.URL)
				r.IsRequired = in.IsRequired
				r.SortOrder = in.SortOrder
			})
			if err != nil {
				return fmt.Errorf("reading %q of %s: %w", in.Title, in.Course, err)
			}
			sum.Readings.record(created, changed)
			keptReadings = append(keptReadings, r.ID)
		}
		if opts.Prune {
			if err := prune(tx, &sum, semesters, courseIDs, keptSessions, keptReadings); err != nil {
				return err
			}
		}
		if opts.DryRun {
			return errDryRun
		}
		return nil
	})
	if err != nil && !errors.Is(err, errDryRun) {
		return SeedSummary{}, err
	}
	return sum, nil
}

// prune deletes what the file no longer lists: courses of its semesters
// (their sessions and readings go with them through the foreign keys), then
// the sessions and readings of the kept courses that the seed did not touch.
func prune(tx *gorm.DB, sum *SeedSummary, semesters map[int]bool, courseIDs map[string]uint, keptSessions, keptReadings []uint) error {
	kept := []uint{0}
	for _, id := range courseIDs {
		kept = append(kept, id)
	}
	sems := []int{}
	for sem := range semesters {
		sems = append(sems, sem)
	}

	var courses []models.Course
	if len(sems) > 0 {
		if err := tx.Where("semester IN ? AND id NOT IN ?", sems, kept).Order("id").Find(&courses).Error; err != nil {
			return err
		}
	}
	for _, c := range courses {
		var sessions, readings int64
		tx.Model(&models.ClassSession{}).Where("course_id = ?", c.ID).Count(&sessions)
		tx.Model(&models.Reading{}).Where("course_id = ?", c.ID).Count(&readings)
		sum.Removed = append(sum.Removed, fmt.Sprintf("course %q (%s, semester %d) with %d sessions and %d readings", c.Name, c.Form, c.Semester, sessions, readings))
		if err := tx.Delete(&models.Course{}, c.ID).Error; err != nil {
			return fmt.Errorf("prune course %q: %w", c.Name, err)
		}
		sum.Courses.Deleted++
		sum.ClassSessions.Deleted += int(sessions)
		sum.Readings.Deleted += int(readings)
	}

	var sessions []models.ClassSession
	if err := tx.Where("course_id IN ? AND id NOT IN ?", kept, keptSessions).Order("starts_at").Find(&sessions).Error; err != nil {
		return err
	}
	for _, s := range sessions {
		sum.Removed = append(sum.Removed, fmt.Sprintf("class session of course %d at %s", s.CourseID, models.FormatDateTime(s.StartsAt)))
		if err := tx.Delete(&models.ClassSession{}, s.ID).Error; err != nil {
			return fmt.Errorf("prune class session %d: %w", s.ID, err)
		}
		sum.ClassSessions.Deleted++
	}

	var readings []models.Reading
	if err := tx.Where("course_id IN ? AND id NOT IN ?", kept, keptReadings).Order("id").Find(&readings).Error; err != nil {
		return err
	}
	for _, r := range readings {
		sum.Removed = append(sum.Removed, fmt.Sprintf("reading %q of course %d", r.Title, r.CourseID))
		if err := tx.Delete(&models.Reading{}, r.ID).Error; err != nil {
			return fmt.Errorf("prune reading %d: %w", r.ID, err)
		}
		sum.Readings.Deleted++
	}
	return nil
}

// nz returns nil for a missing or blank optional text, as the models'
// BeforeSave hooks store it, so an unchanged row compares equal.
func nz(s *string) *string {
	if s == nil || strings.TrimSpace(*s) == "" {
		return nil
	}
	return s
}

// upsert loads the row matching where into row, applies set and writes it:
// an insert when there was no row, an update when set changed a column, and
// nothing otherwise. The comparison is on the row's own fields, before and
// after set, as JSON.
// setInstant sets *dst to t unless it already holds the same instant, so a
// time read back from the database in another zone does not count as a
// change in upsert.
func setInstant(dst *time.Time, t time.Time) {
	if !dst.Equal(t) {
		*dst = t
	}
}

func upsert[T any](tx *gorm.DB, row *T, where clause.Expression, set func(*T)) (created, changed bool, err error) {
	res := tx.Clauses(where).Limit(1).Find(row)
	if res.Error != nil {
		return false, false, res.Error
	}
	if res.RowsAffected == 0 {
		set(row)
		return true, false, tx.Create(row).Error
	}
	before, err := json.Marshal(row)
	if err != nil {
		return false, false, err
	}
	set(row)
	after, err := json.Marshal(row)
	if err != nil {
		return false, false, err
	}
	if bytes.Equal(before, after) {
		return false, false, nil
	}
	return false, true, tx.Save(row).Error
}