honeyDueAPI/cmd/backfill-completion-columns/main.go

// backfill-completion-columns is a one-time migration script that retroactively
// sets the completed_from_column field on existing TaskCompletion records.
//
// Strategy:
//   - One-time tasks: compare completed_at vs due_date to determine if overdue/due_soon/upcoming
//   - Recurring tasks: work backwards from next_due_date using frequency interval to estimate
//     the due date that was active at completion time
//   - Tasks with no due date: default to "completed_tasks" (no way to determine)
//
// Usage:
//   go run ./cmd/backfill-completion-columns --dry-run    # Preview changes
//   go run ./cmd/backfill-completion-columns              # Apply changes
//
// Safe to run multiple times — only updates rows still set to the default "completed_tasks".
package main

import (
	"flag"
	"os"
	"sort"
	"time"

	"github.com/rs/zerolog"
	"github.com/rs/zerolog/log"
	"gorm.io/driver/postgres"
	"gorm.io/gorm"
	"gorm.io/gorm/logger"

	"github.com/treytartt/honeydue-api/internal/config"
)

const daysThreshold = 30 // "due soon" window in days, matches app default

func main() {
	dryRun := flag.Bool("dry-run", false, "Preview changes without writing to database")
	flag.Parse()

	log.Logger = log.Output(zerolog.ConsoleWriter{Out: os.Stderr, TimeFormat: time.RFC3339})

	cfg, err := config.Load()
	if err != nil {
		log.Fatal().Err(err).Msg("Failed to load config")
	}

	dsn := cfg.Database.DSN()

	db, err := gorm.Open(postgres.Open(dsn), &gorm.Config{
		Logger: logger.Default.LogMode(logger.Silent),
	})
	if err != nil {
		log.Fatal().Err(err).Msg("Failed to connect to database")
	}

	log.Info().Bool("dry_run", *dryRun).Msg("Starting backfill of completed_from_column")

	// Query all completions that still have the default value
	type completionRow struct {
		ID          uint
		TaskID      uint
		CompletedAt time.Time
	}
	var completions []completionRow
	err = db.Table("task_taskcompletion").
		Select("id, task_id, completed_at").
		Where("completed_from_column = ? OR completed_from_column IS NULL OR completed_from_column = ''", "completed_tasks").
		Order("task_id, completed_at ASC").
		Scan(&completions).Error
	if err != nil {
		log.Fatal().Err(err).Msg("Failed to query completions")
	}

	log.Info().Int("total_completions", len(completions)).Msg("Found completions to backfill")

	if len(completions) == 0 {
		log.Info().Msg("Nothing to backfill")
		return
	}

	// Load all tasks referenced by these completions
	taskIDs := make(map[uint]bool)
	for _, c := range completions {
		taskIDs[c.TaskID] = true
	}
	taskIDList := make([]uint, 0, len(taskIDs))
	for id := range taskIDs {
		taskIDList = append(taskIDList, id)
	}

	type taskRow struct {
		ID                 uint
		DueDate            *time.Time
		NextDueDate        *time.Time
		FrequencyID        *uint
		CustomIntervalDays *int
		IsCancelled        bool
		IsArchived         bool
	}
	var tasks []taskRow
	err = db.Table("task_task").
		Select("id, due_date, next_due_date, frequency_id, custom_interval_days, is_cancelled, is_archived").
		Where("id IN ?", taskIDList).
		Scan(&tasks).Error
	if err != nil {
		log.Fatal().Err(err).Msg("Failed to query tasks")
	}

	taskMap := make(map[uint]*taskRow, len(tasks))
	for i := range tasks {
		taskMap[tasks[i].ID] = &tasks[i]
	}

	// Load frequency intervals
	type freqRow struct {
		ID   uint
		Name string
		Days *int
	}
	var frequencies []freqRow
	err = db.Table("task_taskfrequency").Select("id, name, days").Scan(&frequencies).Error
	if err != nil {
		log.Fatal().Err(err).Msg("Failed to query frequencies")
	}
	freqMap := make(map[uint]*freqRow, len(frequencies))
	for i := range frequencies {
		freqMap[frequencies[i].ID] = &frequencies[i]
	}

	// Group completions by task for recurring task logic
	taskCompletions := make(map[uint][]completionRow)
	for _, c := range completions {
		taskCompletions[c.TaskID] = append(taskCompletions[c.TaskID], c)
	}
	// Sort each group by completed_at ascending
	for taskID := range taskCompletions {
		sort.Slice(taskCompletions[taskID], func(i, j int) bool {
			return taskCompletions[taskID][i].CompletedAt.Before(taskCompletions[taskID][j].CompletedAt)
		})
	}

	// Determine the column for each completion
	type updateEntry struct {
		completionID uint
		column       string
	}
	var updates []updateEntry

	stats := map[string]int{
		"overdue_tasks":     0,
		"due_soon_tasks":    0,
		"upcoming_tasks":    0,
		"in_progress_tasks": 0,
		"completed_tasks":   0,
		"no_due_date":       0,
	}

	for taskID, comps := range taskCompletions {
		task, ok := taskMap[taskID]
		if !ok {
			continue
		}

		// Get interval days for this task
		var intervalDays int
		isRecurring := false
		if task.FrequencyID != nil {
			if freq, ok := freqMap[*task.FrequencyID]; ok {
				if freq.Name == "Custom" && task.CustomIntervalDays != nil {
					intervalDays = *task.CustomIntervalDays
				} else if freq.Days != nil {
					intervalDays = *freq.Days
				}
			}
		}
		isRecurring = intervalDays > 0

		if !isRecurring {
			// One-time task: compare each completion against due_date
			effectiveDate := task.DueDate
			if effectiveDate == nil {
				// No due date — can't determine, keep default
				for _, c := range comps {
					stats["no_due_date"]++
					updates = append(updates, updateEntry{c.ID, "completed_tasks"})
				}
				continue
			}

			for _, c := range comps {
				column := classifyCompletion(c.CompletedAt, *effectiveDate, daysThreshold)
				stats[column]++
				updates = append(updates, updateEntry{c.ID, column})
			}
		} else {
			// Recurring task: reconstruct due dates by walking backwards from current state
			//
			// Logic: For each completion (oldest first), we estimate what the due date was.
			// - The original DueDate is the first due date
			// - After each completion, the next due date = completedAt + intervalDays
			//
			// So: effectiveDate starts at DueDate, then advances per completion.
			effectiveDate := task.DueDate
			if effectiveDate == nil && task.NextDueDate != nil {
				// If no original DueDate but NextDueDate exists, work backwards
				// from the most recent completion
				est := task.NextDueDate.AddDate(0, 0, -intervalDays*len(comps))
				effectiveDate = &est
			}
			if effectiveDate == nil {
				for _, c := range comps {
					stats["no_due_date"]++
					updates = append(updates, updateEntry{c.ID, "completed_tasks"})
				}
				continue
			}

			currentDueDate := *effectiveDate
			for _, c := range comps {
				column := classifyCompletion(c.CompletedAt, currentDueDate, daysThreshold)
				stats[column]++
				updates = append(updates, updateEntry{c.ID, column})

				// Advance due date for next completion
				currentDueDate = c.CompletedAt.AddDate(0, 0, intervalDays)
			}
		}
	}

	// Log stats
	log.Info().
		Int("overdue", stats["overdue_tasks"]).
		Int("due_soon", stats["due_soon_tasks"]).
		Int("upcoming", stats["upcoming_tasks"]).
		Int("no_due_date", stats["no_due_date"]).
		Int("total_updates", len(updates)).
		Msg("Classification results")

	if *dryRun {
		log.Info().Msg("Dry run complete — no changes written")
		return
	}

	// Apply updates in batches
	batchSize := 500
	updated := 0
	for i := 0; i < len(updates); i += batchSize {
		end := i + batchSize
		if end > len(updates) {
			end = len(updates)
		}
		batch := updates[i:end]

		err := db.Transaction(func(tx *gorm.DB) error {
			for _, u := range batch {
				if err := tx.Table("task_taskcompletion").
					Where("id = ?", u.completionID).
					Update("completed_from_column", u.column).Error; err != nil {
					return err
				}
			}
			return nil
		})
		if err != nil {
			log.Fatal().Err(err).Int("batch_start", i).Msg("Failed to update batch")
		}
		updated += len(batch)
		log.Info().Int("updated", updated).Int("total", len(updates)).Msg("Progress")
	}

	log.Info().Int("total_updated", updated).Msg("Backfill complete")
}

// classifyCompletion determines what kanban column a task was in when completed.
func classifyCompletion(completedAt time.Time, dueDate time.Time, threshold int) string {
	// Normalize to date-only comparison (matches app behavior)
	completedDay := time.Date(completedAt.Year(), completedAt.Month(), completedAt.Day(), 0, 0, 0, 0, time.UTC)
	dueDay := time.Date(dueDate.Year(), dueDate.Month(), dueDate.Day(), 0, 0, 0, 0, time.UTC)

	if completedDay.After(dueDay) {
		return "overdue_tasks"
	}

	daysBefore := dueDay.Sub(completedDay).Hours() / 24
	if daysBefore <= float64(threshold) {
		return "due_soon_tasks"
	}

	return "upcoming_tasks"
}