tests/test_learning_engine.py

"""Tests for Course Builder and Exam Coach Session services."""
from __future__ import annotations

from app.services.course_builder import (
    build_course_plan,
    course_plan_to_output,
    _detect_course_type,
)
from app.services.exam_coach_session import (
    build_exam_coach_session,
    exam_coach_session_to_output,
    _parse_daily_time,
    _compute_exam_countdown,
)
from app.services.pyq_pattern_analyzer import PatternAnalysisResult, TopicPattern


def test_learning_engine_routes_registered(client):
    course_response = client.post(
        "/generate/course-plan",
        json={"raw_request": "Teach me machine learning"},
    )
    assert course_response.status_code == 200, course_response.text
    assert course_response.json()["lessons_total"] > 0

    session_response = client.post(
        "/generate/exam-coach-session",
        json={
            "subject": "Physics",
            "topic": "Electric Charges",
            "daily_time": "30 min",
        },
    )
    assert session_response.status_code == 200, session_response.text
    assert session_response.json()["study_blocks"]


def test_learning_engine_routes_require_auth(auth_client):
    response = auth_client.post(
        "/generate/course-plan",
        json={"raw_request": "Teach me machine learning"},
    )
    assert response.status_code in {401, 403}


# === Course Builder Tests ===

def test_course_plan_disaster_management_semester():
    result = build_course_plan(
        raw_request="Teach me disaster management 4th semester degree",
        semester="4",
        degree="degree",
    )
    assert result.title != ""
    assert result.lessons_total > 0
    assert result.estimated_total_minutes > 0
    assert len(result.modules) > 0
    assert result.data_source_label in {"Starter seed", "Uploaded source"}
    assert result.confidence > 0


def test_course_plan_machine_learning():
    result = build_course_plan(
        raw_request="Teach me machine learning",
    )
    output = course_plan_to_output(result)
    assert output["title"] != ""
    assert output["lessons_total"] > 0
    assert len(output["modules"]) > 0
    assert "Machine Learning" in output["subject_area"] or "Machine Learning" in output["title"]
    assert len(output["prerequisites"]) > 0
    assert len(output["practice_tasks"]) > 0
    assert len(output["quizzes"]) > 0


def test_course_plan_with_source_context():
    result = build_course_plan(
        raw_request="Teach me Physics chapter 1",
        source_context="Electric charges and Coulomb's law are fundamental concepts...",
        subject="Physics",
        topic="Electric Charges and Fields",
    )
    assert result.source_basis == "source_upload"
    assert result.data_source_label == "Uploaded source"


def test_course_plan_source_ids_without_context_not_marked_uploaded():
    result = build_course_plan(
        raw_request="Teach me Physics chapter 1",
        source_ids=["missing-or-not-ready-source"],
        subject="Physics",
        topic="Electric Charges and Fields",
    )
    assert result.source_basis == "catalog_seed"
    assert result.data_source_label == "Starter seed"
    assert any("no usable source text" in note.lower() for note in result.trust_notes)


def test_course_plan_no_source_uses_starter_seed():
    result = build_course_plan(
        raw_request="Teach me Chemistry mole concept",
        subject="Chemistry",
        topic="Mole Concept",
    )
    assert result.source_basis == "catalog_seed"
    assert result.data_source_label == "Starter seed"
    assert any("No source" in note for note in result.trust_notes)


def test_course_plan_degree_semester():
    result = build_course_plan(
        raw_request="Teach me disaster management for 4th semester B.Tech",
        semester="4",
        degree="B.Tech",
    )
    assert len(result.modules) >= 3
    assert result.lessons_total >= 8


def test_course_plan_skill_course():
    result = build_course_plan(
        raw_request="Learn Python programming from scratch",
    )
    assert len(result.modules) >= 3
    assert any("Python" in m.title for m in result.modules)


def test_course_plan_playlist_metadata():
    result = build_course_plan(
        raw_request="Turn this ML playlist into a course",
        playlist_metadata={
            "topics": [
                "Linear Regression",
                {"title": "Gradient Descent"},
                "Model Evaluation",
            ],
        },
    )
    assert result.learner_level == "Playlist"
    assert result.lessons_total > 0
    assert any("Linear Regression" in module.title for module in result.modules)


def test_course_plan_confidence_increases_with_info():
    low = build_course_plan(raw_request="Teach something")
    high = build_course_plan(
        raw_request="Teach me +2 Physics chapter 1",
        subject="Physics",
        topic="Electric Charges",
        class_level="+2",
    )
    assert high.confidence > low.confidence


def test_course_plan_output_json_serializable():
    result = build_course_plan(
        raw_request="Teach me Physics Ray Optics",
        subject="Physics",
        topic="Ray Optics",
    )
    output = course_plan_to_output(result)
    assert isinstance(output, dict)
    assert "modules" in output
    assert "quizzes" in output
    assert "practice_tasks" in output
    assert isinstance(output["modules"], list)


def test_course_plan_school_physics_chapter():
    result = build_course_plan(
        raw_request="Teach me +2 Physics Ray Optics",
        subject="Physics",
        topic="Ray Optics",
    )
    assert result.subject_area == "Physics"
    assert len(result.modules) > 0
    assert result.lessons_total > 0


def test_course_plan_school_tuition_contract_contains_required_sections():
    result = build_course_plan(
        raw_request="Teach me lens maker formula",
        class_level="Plus Two / Grade 12",
        syllabus="Kerala HSE",
        subject="Physics",
        chapter="Ray Optics",
        exam_date="2026-07-10",
        daily_study_time="1 hour",
        goal="board-exam",
    )
    output = course_plan_to_output(result)

    assert output["student_context"]["class_level"] == "Plus Two / Grade 12"
    assert output["student_context"]["syllabus"] == "Kerala HSE"
    assert output["student_context"]["subject"] == "Physics"
    assert output["student_context"]["chapter"] == "Ray Optics"
    assert output["student_context"]["exam_date"] == "2026-07-10"
    assert output["student_context"]["daily_study_time"] == "1 hour"
    assert output["lesson_outline"]
    assert output["prerequisite_check"]
    assert output["concept_explanation"]["explanation"]
    assert output["derivation_or_problem_steps"][0]["steps"]
    assert output["practice_questions"]
    assert output["pyq_style_questions"]
    assert output["weak_topic_repairs"]
    assert output["revision_plan"]
    assert output["estimated_study_time"]


def test_course_plan_route_accepts_full_school_context(client):
    response = client.post(
        "/generate/course-plan",
        json={
            "raw_request": "Teach me Boolean logic",
            "class_level": "Class 11",
            "syllabus": "Kerala HSE",
            "subject": "Computer Science",
            "chapter": "Boolean Logic",
            "exam_date": "2026-07-10",
            "daily_study_time": "30 minutes",
            "current_level": "basic",
            "weak_topics": ["logic gates"],
            "goal": "concept-clarity",
        },
    )

    assert response.status_code == 200, response.text
    data = response.json()
    assert data["student_context"]["syllabus"] == "Kerala HSE"
    assert data["student_context"]["chapter"] == "Boolean Logic"
    assert data["student_context"]["daily_study_time"] == "30 minutes"
    assert data["lesson_outline"]
    assert data["practice_questions"]
    assert data["revision_plan"]
    assert "Student self-reported level: basic." in data["trust_notes"]
    assert data["weak_topic_repairs"][0]["topic"] == "logic gates"


# === Detect Course Type Tests ===

def test_detect_course_type_degree():
    assert _detect_course_type("4th semester degree") == "degree"
    assert _detect_course_type("B.Tech computer science") == "degree"
    assert _detect_course_type("MCA 2nd sem") == "degree"


def test_detect_course_type_skill():
    assert _detect_course_type("Learn machine learning") == "skill"
    assert _detect_course_type("Course on Python") == "skill"


def test_detect_course_type_topic():
    assert _detect_course_type("Teach me Physics chapter 1") == "topic"


# === Exam Coach Session Tests ===

def test_exam_coach_session_plus2_physics_1hour():
    result = build_exam_coach_session(
        class_level="+2",
        board="Kerala HSE",
        subject="Physics",
        chapter="Electric Charges and Fields",
        daily_time="1 hour",
    )
    assert result.session_title != ""
    assert result.time_budget == "1 hour"
    assert len(result.study_blocks) > 0
    assert result.today_mission != ""
    assert result.concept_teaching_task != {}
    assert result.answer_writing_task != {}
    assert result.revision_task != {}
    assert result.quick_test != {}


def test_exam_coach_session_1_5_hour_derivation():
    result = build_exam_coach_session(
        class_level="+2",
        subject="Physics",
        topic="Equations of Motion",
        daily_time="1.5 hour",
    )
    assert result.time_budget == "1.5 hour"
    assert len(result.study_blocks) == 3
    assert result.derivation_or_numerical_task.get("task_type") == "derivation"


def test_exam_coach_session_no_pyq_no_fake():
    result = build_exam_coach_session(
        subject="Physics",
        topic="Electric Charges",
        daily_time="1 hour",
    )
    pyq_task = result.pyq_task
    assert pyq_task["has_evidence"] is False
    assert "No uploaded PYQ" in pyq_task["description"] or "No PYQ" in pyq_task["data_source_label"]
    assert any("No PYQ" in note for note in result.trust_notes)


def test_exam_coach_session_plus2_motion_does_not_leak_unrelated_seed():
    result = build_exam_coach_session(
        class_level="+2",
        subject="Physics",
        topic="+2 Physics motion in a straight line",
        daily_time="1 hour",
        exam_date="in 20 days",
    )
    concept_output = result.concept_teaching_task["output"]
    rendered = str(concept_output).lower()

    assert concept_output["chapter_title"] == "+2 Physics motion in a straight line"
    assert concept_output["data_source_label"] == "Starter seed"
    assert "electric charges" not in rendered
    assert "coulomb" not in rendered
    assert "gauss" not in rendered
    assert result.pyq_task["has_evidence"] is False
    assert result.data_source_label == "Starter seed"


def test_exam_coach_session_chemistry_numerical_focus():
    result = build_exam_coach_session(
        subject="Chemistry",
        topic="Chemistry mole concept numericals",
        daily_time="1 hour",
    )
    task = result.derivation_or_numerical_task
    concept_output = result.concept_teaching_task["output"]

    assert task["task_type"] == "numerical"
    assert "numerical" in task["description"].lower()
    assert "Chemistry mole concept numericals" == concept_output["chapter_title"]
    assert "given data" in concept_output["board_answer_keywords"]
    assert result.data_source_label == "Starter seed"


def test_exam_coach_session_maths_proof_focus():
    result = build_exam_coach_session(
        subject="Maths",
        topic="Maths trigonometry proof revision",
        daily_time="1 hour",
    )
    task = result.derivation_or_numerical_task
    concept_output = result.concept_teaching_task["output"]

    assert task["task_type"] == "proof"
    assert "proof" in task["title"].lower()
    assert "given" in concept_output["board_answer_keywords"]
    assert "to prove" in concept_output["board_answer_keywords"]
    assert "reason" in concept_output["board_answer_keywords"]


def test_exam_coach_session_unknown_topic_uses_requested_starter_seed():
    result = build_exam_coach_session(
        subject="Physics",
        topic="Quantum banana curves",
        daily_time="30 min",
    )
    concept_output = result.concept_teaching_task["output"]
    rendered = str(concept_output).lower()

    assert concept_output["chapter_title"] == "Quantum banana curves"
    assert concept_output["data_source_label"] == "Starter seed"
    assert "electric charges" not in rendered
    assert "ray optics" not in rendered
    assert "coulomb" not in rendered
    assert result.pyq_task["has_evidence"] is False


def test_exam_coach_session_with_pyq_evidence():
    pyq_data = [
        {"question_text": "State Coulomb's law", "marks": 3, "year": "2023"},
        {"question_text": "Derive electric field", "marks": 5, "year": "2022"},
    ]
    result = build_exam_coach_session(
        subject="Physics",
        topic="Electric Charges",
        daily_time="1 hour",
        pyq_data=pyq_data,
    )
    pyq_task = result.pyq_task
    assert pyq_task["has_evidence"] is True
    assert "Uploaded PYQs" in pyq_task["data_source_label"]


def test_exam_coach_session_unknown_exam_date():
    result = build_exam_coach_session(
        subject="Physics",
        topic="Electric Charges",
        daily_time="1 hour",
        exam_date=None,
    )
    assert result.exam_countdown == "unknown"
    assert any("Exam date not provided" in note for note in result.trust_notes)


def test_exam_coach_session_exam_tomorrow():
    result = build_exam_coach_session(
        subject="Physics",
        topic="Electric Charges",
        daily_time="1 hour",
        exam_date="tomorrow",
    )
    assert "tomorrow" in result.exam_countdown.lower()


def test_exam_coach_session_30_min():
    result = build_exam_coach_session(
        subject="Physics",
        topic="Electric Charges",
        daily_time="30 min",
    )
    assert result.time_budget == "30 min"
    assert len(result.study_blocks) == 1
    assert result.study_blocks[0].duration_minutes == 30


def test_exam_coach_session_weak_topics():
    result = build_exam_coach_session(
        subject="Physics",
        topic="Electric Charges",
        daily_time="1 hour",
        weak_topics=["Electromagnetic Induction", "Current Electricity"],
    )
    revision = result.revision_task
    assert "Electromagnetic Induction" in revision.get("revision_focus", [])
    assert "Current Electricity" in revision.get("revision_focus", [])


def test_exam_coach_session_with_source():
    result = build_exam_coach_session(
        subject="Physics",
        topic="Electric Charges",
        daily_time="1 hour",
        source_context="Electric charge is a fundamental property of matter...",
    )
    assert result.data_source_label == "Uploaded source"
    assert not any("No source" in note for note in result.trust_notes)


def test_exam_coach_session_source_ids_without_context_not_marked_uploaded():
    result = build_exam_coach_session(
        subject="Physics",
        topic="Electric Charges",
        daily_time="1 hour",
        source_ids=["missing-or-not-ready-source"],
    )
    assert result.data_source_label == "Starter seed"
    assert any("no usable source text" in note.lower() for note in result.trust_notes)


def test_exam_coach_session_last_night_mode():
    result = build_exam_coach_session(
        subject="Physics",
        topic="Electric Charges",
        daily_time="last-night",
    )
    assert result.time_budget == "last-night"
    assert "Last-night revision" in result.today_mission
    assert len(result.study_blocks) == 3


def test_exam_coach_session_avoids_unsupported_high_yield_claims():
    result = build_exam_coach_session(
        subject="Physics",
        topic="Electric Charges",
        daily_time="30 min",
        exam_date="tomorrow",
    )
    rendered = str(exam_coach_session_to_output(result)).lower()
    assert "high-yield" not in rendered
    assert "sure shot" not in rendered
    assert "guaranteed" not in rendered


def test_exam_coach_session_output_json_serializable():
    result = build_exam_coach_session(
        subject="Physics",
        topic="Electric Charges",
        daily_time="1 hour",
    )
    output = exam_coach_session_to_output(result)
    assert isinstance(output, dict)
    assert "study_blocks" in output
    assert "concept_teaching_task" in output
    assert "quick_test" in output


def test_exam_coach_session_next_day_plan():
    result = build_exam_coach_session(
        subject="Physics",
        topic="Electric Charges",
        daily_time="1 hour",
        weak_topics=["Magnetism"],
    )
    next_plan = result.next_day_plan
    assert "Magnetism" in str(next_plan.get("suggested_topics", []))


# === Parse Daily Time Tests ===

def test_parse_daily_time_30_min():
    label, minutes = _parse_daily_time("30 min")
    assert minutes == 30


def test_parse_daily_time_1_hour():
    label, minutes = _parse_daily_time("1 hour")
    assert minutes == 60


def test_parse_daily_time_1_5_hour():
    label, minutes = _parse_daily_time("1.5 hour")
    assert minutes == 90


def test_parse_daily_time_last_night():
    label, minutes = _parse_daily_time("last-night")
    assert label == "last-night"
    assert minutes == 90


def test_parse_daily_time_2_hours():
    label, minutes = _parse_daily_time("2 hours")
    assert minutes == 120


# === Compute Exam Countdown Tests ===

def test_compute_exam_countdown_days():
    label, days = _compute_exam_countdown("3 days")
    assert days == 3
    assert "3" in label


def test_compute_exam_countdown_weeks():
    label, days = _compute_exam_countdown("2 weeks")
    assert days == 14


def test_compute_exam_countdown_unknown():
    label, days = _compute_exam_countdown(None)
    assert label == "unknown"
    assert days is None


def test_compute_exam_countdown_tomorrow():
    label, days = _compute_exam_countdown("tomorrow")
    assert days == 1


# === Integration: Course + Exam Session ===

def test_course_plan_and_exam_session_same_subject():
    course = build_course_plan(
        raw_request="Teach me +2 Physics",
        subject="Physics",
    )
    session = build_exam_coach_session(
        subject="Physics",
        daily_time="1 hour",
    )
    assert course.subject_area == session.concept_teaching_task.get("subject", "Physics")