from django.conf import settings
from django.db import models

class Course(models.Model):
    title = models.CharField(max_length=200)

class Lesson(models.Model):
    course = models.ForeignKey(Course, on_delete=models.CASCADE, related_name="lessons")
    title = models.CharField(max_length=200)
    order = models.PositiveIntegerField()

class LessonProgress(models.Model):
    user = models.ForeignKey(settings.AUTH_USER_MODEL, on_delete=models.CASCADE)
    lesson = models.ForeignKey(Lesson, on_delete=models.CASCADE)
    completed_at = models.DateTimeField(auto_now_add=True)

    class Meta:
        unique_together = ("user", "lesson")   # no duplicate completions

from fastapi import FastAPI

app = FastAPI()
progress = {}   # (user_id, course_id) -> set of lesson ids; use a database in real apps

@app.post("/progress/{user_id}/{course_id}/{lesson_id}")
def complete(user_id: int, course_id: int, lesson_id: int):
    progress.setdefault((user_id, course_id), set()).add(lesson_id)
    return {"completed": len(progress[(user_id, course_id)])}

@app.get("/progress/{user_id}/{course_id}")
def get_progress(user_id: int, course_id: int, total_lessons: int = 20):
    done = len(progress.get((user_id, course_id), set()))
    return {"percent": round(100 * done / total_lessons, 1)}

import pandas as pd

events = pd.read_csv("lesson_events.csv")   # user_id, lesson_order, event ('start'/'complete')
completed = events[events["event"] == "complete"]
by_lesson = completed.groupby("lesson_order")["user_id"].nunique()
drop = (1 - by_lesson / by_lesson.shift(1)).round(3)
print(pd.DataFrame({"learners": by_lesson, "drop_from_prev": drop}).head(10))
# Large drops point to lessons that may need redesign

week = pd.read_csv("learner_week.csv")   # user_id, attended, submitted, due
week["submission_rate"] = week["submitted"] / week["due"].where(week["due"] > 0)
needs_help = week[(week["attended"] < 2) | (week["submission_rate"] < 0.5)]
needs_help.to_csv("mentor_followups.csv", index=False)
print(len(needs_help), "learners flagged for a friendly check-in")

# Simplified idea: check a learner's function against test cases
# Real platforms run untrusted code in isolated sandboxes, never directly.
tests = [((2, 3), 5), ((-1, 1), 0), ((0, 0), 0)]

def learner_add(a, b):          # imagine this came from a submission
    return a + b

results = [learner_add(*args) == expected for args, expected in tests]
print(f"Passed {sum(results)} of {len(tests)} tests")

def next_practice(scores_by_topic, threshold=70):
    """Recommend the weakest topic below threshold, else the next new topic."""
    weak = {t: s for t, s in scores_by_topic.items() if s < threshold}
    if weak:
        return min(weak, key=weak.get)
    return "next_new_topic"

print(next_practice({"loops": 82, "functions": 64, "lists": 58}))   # lists

Project: Mini learning platform
Backend: Django or FastAPI with courses, lessons, enrolments, progress
Auth: learner, mentor and admin roles
Analytics: lesson drop-off report with pandas
Automation: weekly progress summary email (console output in demo)
Extras: quiz with auto-scoring, unique progress constraints
Docs: README, API docs, ER diagram

Edtech backend checklist
☐ Role-based permissions (learner, mentor, admin)
☐ Progress cannot be double-counted
☐ Timezone-aware class schedules
☐ Background jobs for reminders
☐ Privacy: minimal personal data, careful with minors' data
☐ Tests for enrolment and progress logic

Project story
1. Problem: track learner progress reliably
2. Design: models, constraints, APIs
3. Analytics: which lessons lose learners
4. One hard bug and fix
5. What you'd add: recommendations, accessibility