[FIX] completion_rate 100% 초과 방지 및 진행률 계산 개선

[b1351b1]

개요

• 강의 진행률 API에서 completion_rate가 100%를 초과하는 버그 수정

변경사항

1. DB 스키마 변경

• Progress 모델에 unique_watched_seconds 필드 추가
  • watched_seconds: 총 시청 시간 (반복 재생 포함 누적) - 분석용
  • unique_watched_seconds: 유니크 시청 시간 - 진행률 계산용
• Alembic 마이그레이션 파일 생성

2. 진행률 계산 로직 개선

• completion_rate 계산 시 unique_watched_seconds 사용
• completion_rate를 min(completion_rate, 100.0)으로 100% 이하로 제한
• last_position을 duration_seconds 이하로 제한

3. 수정된 파일

• app/models/progress.py: unique_watched_seconds 필드 추가
• app/services/enrollment_service.py: 진행률 계산 로직 전면 개선
• alembic/versions/a3a28d52a24f_add_unique_watched_seconds_to_progress.py:
  마이그레이션 파일

4. 기대 효과

• completion_rate가 항상 0~100% 범위 내에서 반환
• 반복 재생 시에도 정확한 진행률 표시
• 학습 분석(watched_seconds)과 진행률(unique_watched_seconds) 분리로 데이터
  활용도 향상

관련 이슈

**Closes #105

[Copilot]

Pull request overview

This PR addresses a bug where completion_rate could exceed 100% by introducing a new database field unique_watched_seconds to track the unique (non-repeated) watch time for progress calculations, while preserving watched_seconds for analytics purposes. The PR adds proper capping logic to ensure completion rates never exceed 100%.

Key Changes:

• Added unique_watched_seconds field to the Progress model with an Alembic migration
• Updated progress calculation logic to use unique_watched_seconds instead of watched_seconds
• Applied min(completion_rate, 100.0) capping throughout the codebase

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 7 comments.

File	Description
app/models/progress.py	Added unique_watched_seconds field to track unique watch time for progress calculations
app/services/enrollment_service.py	Updated all progress calculation functions to use unique_watched_seconds and apply 100% capping
alembic/versions/a3a28d52a24f_add_unique_watched_seconds_to_progress.py	Database migration to add the new field with default values

Critical Issues Found:

1. Logic Bug: The unique_watched_seconds calculation (= last_position) will cause the value to decrease when users rewind videos, which defeats its purpose. It should use max(existing_value, last_position) to only track forward progress.
2. Migration Data Issue: Existing records are migrated with unique_watched_seconds = watched_seconds, which may still allow >100% completion rates until users update their progress.
3. Documentation Mismatch: Comments suggest a calculation is happening, but the implementation is just a direct assignment that doesn't properly handle the "unique" aspect.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Critical Logic Bug: The unique_watched_seconds calculation is incorrect. Setting it equal to last_position means that if a user watches a video from the beginning multiple times, the unique watched seconds won't update properly.

For example:

• User watches 0-100 seconds → last_position=100, unique_watched_seconds=100 ✓
• User rewinds and watches 0-150 seconds → last_position=150, unique_watched_seconds=150 ✓
• User rewinds to 50 and watches to 120 → last_position=120, unique_watched_seconds=120 ✗ (should still be 150)

The current implementation only tracks the furthest position reached (last_position), which already serves that purpose. unique_watched_seconds should be calculated differently (e.g., tracking unique segments watched) or should be set to max(unique_watched_seconds, last_position) when updating existing records to preserve the maximum progress.

[Copilot]

@b1351b1 I've opened a new pull request, #112, to work on those changes. Once the pull request is ready, I'll request review from you.

[Copilot]

@b1351b1 I've opened a new pull request, #113, to work on those changes. Once the pull request is ready, I'll request review from you.

[b1351b1]

@copilot open a new pull request to apply changes based on the comments in this thread

[Copilot]

@b1351b1 I've opened a new pull request, #114, to work on those changes. Once the pull request is ready, I'll request review from you.