seed_syllabi.py

"""
Seed syllabus structure trees for every subject in the registry.

Why this exists
---------------
The data API ships with a SUBJECT_REGISTRY (24 Cambridge/ZIMSEC subjects) but the
syllabus trees in Firebase were never populated. With no tree, the app server has
nothing to scaffold a learning path on, so every subject showed empty topic lists
and "0 of 0 lessons". Lessons and quizzes are generated by AI at runtime per node,
so a subject only needs a faithful tree of  topic -> [lessons]  to become usable.

This script holds an AI-authored structure for each subject (real syllabus topics
and the sub-lessons under them), turns it into the tree shape the app expects, and
POSTs the whole batch to the data API's /v1/admin/seed-structures endpoint.

Tree node shape consumed by the app:
    { "id": "...", "title": "...", "type": "topic" | "lesson", "children": [...] }

Usage
-----
    python seed_syllabi.py                 # seed the live data API
    DATA_API_BASE_URL=http://localhost:7860/v1 python seed_syllabi.py
    python seed_syllabi.py --dry-run       # print what would be sent, send nothing
"""

import os
import sys
import json
import requests

DATA_API_BASE_URL = os.environ.get(
    "DATA_API_BASE_URL", "https://rairo-marka-data-api.hf.space/v1"
)

# -----------------------------------------------------------------------------
# Authored syllabus content.
#   remote_id -> { topic_title: [lesson_title, ...], ... }
# Topic order is the intended teaching order; lessons sit under each topic.
# -----------------------------------------------------------------------------

SYLLABI = {
    # ===================== A-LEVEL =====================
    "A_9706": {  # Accounting
        "The Accounting Cycle": [
            "Double-entry bookkeeping", "Books of prime entry",
            "The ledger and trial balance", "Accounting concepts and conventions",
        ],
        "Financial Statements": [
            "Income statement", "Statement of financial position",
            "Accruals and prepayments", "Depreciation of non-current assets",
            "Irrecoverable and doubtful debts",
        ],
        "Partnership Accounts": [
            "Appropriation of profit", "Capital and current accounts",
            "Goodwill", "Admission and retirement of partners",
        ],
        "Company Accounts": [
            "Share capital and reserves", "Issue of shares and debentures",
            "Published financial statements", "Statement of cash flows",
        ],
        "Cost and Management Accounting": [
            "Cost classification", "Marginal costing", "Absorption costing",
            "Break-even analysis", "Budgeting and budgetary control",
        ],
        "Analysis and Communication": [
            "Accounting ratios", "Interpretation of financial statements",
        ],
    },
    "A_9700": {  # Biology
        "Cell Structure": [
            "Microscopy and magnification", "Prokaryotic and eukaryotic cells",
            "Cell organelles", "Cell membranes and transport",
        ],
        "Biological Molecules": [
            "Carbohydrates", "Lipids", "Proteins", "Water and inorganic ions", "Enzymes",
        ],
        "Genetic Information": [
            "DNA structure and replication", "Protein synthesis",
            "Mitosis and the cell cycle", "Meiosis and variation",
        ],
        "Transport Systems": [
            "Transport in plants", "Transport in mammals",
            "The mammalian heart", "Haemoglobin and gas transport",
        ],
        "Disease and Immunity": [
            "Infectious diseases", "Antibiotics", "The immune system", "Vaccination",
        ],
        "Energy and Coordination": [
            "Photosynthesis", "Respiration", "Homeostasis", "Nervous coordination",
        ],
    },
    "A_9609": {  # Business
        "Business and its Environment": [
            "Enterprise and entrepreneurs", "Business structure",
            "Size of business", "Business objectives", "Stakeholders in business",
        ],
        "Human Resource Management": [
            "Motivation theories", "Leadership and management styles",
            "Human resource planning", "Organisation structure",
        ],
        "Marketing": [
            "Marketing planning", "Market research", "The marketing mix", "Marketing strategy",
        ],
        "Operations Management": [
            "Operations planning", "Inventory management",
            "Capacity utilisation", "Quality management", "Lean production",
        ],
        "Finance and Accounting": [
            "Sources of finance", "Costs and break-even",
            "Budgets and variances", "Investment appraisal",
        ],
        "Strategic Management": [
            "Strategic analysis (SWOT and PEST)", "Strategic choice", "Managing change",
        ],
    },
    "A_9701": {  # Chemistry
        "Atomic Structure": [
            "Atoms and isotopes", "Electronic configuration",
            "Ionisation energies", "Mass spectrometry",
        ],
        "Chemical Bonding": [
            "Ionic bonding", "Covalent bonding", "Shapes of molecules",
            "Intermolecular forces", "Metallic bonding",
        ],
        "Chemical Energetics": [
            "Enthalpy changes", "Hess's law", "Bond energies", "Born-Haber cycles",
        ],
        "Equilibria": [
            "Reversible reactions", "Kc and Kp", "Le Chatelier's principle",
            "Acids and bases", "pH and buffer calculations",
        ],
        "Reaction Kinetics": [
            "Rates of reaction", "Rate equations", "Reaction mechanisms", "Catalysis",
        ],
        "Organic Chemistry": [
            "Hydrocarbons", "Halogenoalkanes", "Alcohols and esters",
            "Carbonyl compounds", "Organic analysis",
        ],
    },
    "A_9618": {  # Computer Science
        "Information Representation": [
            "Number systems", "Binary and hexadecimal", "Data types",
            "Character sets and image representation",
        ],
        "Communication and Networking": [
            "Networks and topologies", "Internet technologies", "Protocols",
        ],
        "Hardware and Processors": [
            "Logic gates", "Boolean algebra", "CPU architecture", "Assembly language",
        ],
        "System Software": [
            "Operating systems", "Language translators", "Memory management",
        ],
        "Databases": [
            "Relational databases", "Normalisation", "SQL", "Database management systems",
        ],
        "Algorithms and Programming": [
            "Programming paradigms", "Data structures", "Searching and sorting",
            "Recursion", "Abstract data types",
        ],
    },
    "A_9708": {  # Economics
        "Basic Economic Ideas": [
            "Scarcity and choice", "Factors of production",
            "Production possibility curves", "Specialisation and exchange",
        ],
        "The Price System": [
            "Demand and supply", "Elasticities", "Market equilibrium",
            "Consumer and producer surplus",
        ],
        "Market Failure": [
            "Externalities", "Public goods", "Information failure",
            "Government intervention",
        ],
        "The Macroeconomy": [
            "Aggregate demand and supply", "National income", "Economic growth",
            "Inflation and unemployment",
        ],
        "Government Macro Policy": [
            "Fiscal policy", "Monetary policy", "Supply-side policy",
            "Balance of payments",
        ],
        "International Economics": [
            "Exchange rates", "Trade and protectionism", "Globalisation",
        ],
    },
    "A_9231": {  # Further Mathematics
        "Further Pure 1": [
            "Roots of polynomial equations", "Rational functions and graphs",
            "Summation of series", "Matrices", "Polar coordinates",
        ],
        "Further Pure 2": [
            "Hyperbolic functions", "Further integration",
            "Complex numbers", "Differential equations",
        ],
        "Further Mechanics": [
            "Momentum and impulse", "Circular motion",
            "Work, energy and power", "Simple harmonic motion",
        ],
        "Further Probability and Statistics": [
            "Continuous random variables", "Poisson distribution",
            "Hypothesis testing", "Chi-squared tests",
        ],
    },
    "A_9489": {  # History
        "Modern Europe 1750-1921": [
            "The French Revolution", "The Industrial Revolution",
            "Nationalism and unification", "The Russian Revolution",
        ],
        "International Relations 1871-1945": [
            "Origins of the First World War", "The Treaty of Versailles",
            "Rise of dictatorships", "Origins of the Second World War",
        ],
        "The Cold War": [
            "Origins of the Cold War", "Berlin and Cuba crises",
            "Détente", "The end of the Cold War",
        ],
        "African History": [
            "Colonialism in Africa", "African nationalism",
            "Decolonisation", "Post-independence Africa",
        ],
    },
    "A_9695": {  # Literature in English
        "Poetry": [
            "Reading and responding to poetry", "Form and structure",
            "Imagery and language", "Comparative analysis",
        ],
        "Prose": [
            "The novel and narrative voice", "Characterisation",
            "Themes and context", "Critical interpretation",
        ],
        "Drama": [
            "Dramatic techniques", "Tragedy", "Comedy", "Staging and performance",
        ],
        "Shakespeare": [
            "Shakespearean language", "Themes and ideas",
            "Character study", "Critical perspectives",
        ],
        "Critical Skills": [
            "Unseen criticism", "Contextual analysis", "Essay writing",
        ],
    },
    "A_9709": {  # Mathematics
        "Pure Mathematics 1": [
            "Quadratics", "Functions", "Coordinate geometry", "Circular measure",
            "Trigonometry", "Series", "Differentiation", "Integration",
        ],
        "Pure Mathematics 3": [
            "Algebra", "Logarithmic and exponential functions",
            "Trigonometry", "Differentiation and integration",
            "Numerical solution of equations", "Vectors",
            "Differential equations", "Complex numbers",
        ],
        "Mechanics": [
            "Forces and equilibrium", "Kinematics of motion",
            "Newton's laws of motion", "Energy, work and power", "Momentum",
        ],
        "Probability and Statistics": [
            "Representation of data", "Permutations and combinations",
            "Probability", "Discrete random variables", "The normal distribution",
        ],
    },
    "A_9702": {  # Physics
        "Mechanics": [
            "Kinematics", "Dynamics", "Forces and equilibrium",
            "Work, energy and power", "Momentum",
        ],
        "Matter and Materials": [
            "Density and pressure", "Deformation of solids", "Phases of matter",
        ],
        "Oscillations and Waves": [
            "Wave properties", "Superposition", "Stationary waves",
            "Simple harmonic motion", "The Doppler effect",
        ],
        "Electricity and Magnetism": [
            "Electric fields", "Current electricity", "D.C. circuits",
            "Magnetic fields", "Electromagnetic induction",
        ],
        "Modern Physics": [
            "Quantum physics", "The photoelectric effect",
            "Atomic and nuclear physics", "Radioactivity",
        ],
        "Thermal Physics": [
            "Temperature and thermal energy", "Ideal gases", "Thermodynamics",
        ],
    },
    "A_9699": {  # Sociology
        "Socialisation and Identity": [
            "Culture and identity", "Socialisation",
            "Social control", "Nature versus nurture",
        ],
        "Sociological Theory": [
            "Functionalism", "Marxism", "Feminism", "Social action theory",
        ],
        "Family": [
            "Family forms and diversity", "Changing family roles",
            "Marriage and divorce", "Childhood",
        ],
        "Education": [
            "The role of education", "Educational achievement",
            "The hidden curriculum", "Education policy",
        ],
        "Research Methods": [
            "Quantitative methods", "Qualitative methods",
            "Sampling", "Reliability and validity",
        ],
        "Social Inequality": [
            "Class inequality", "Gender inequality", "Ethnicity", "Poverty",
        ],
    },
    "A_9395": {  # Travel & Tourism
        "The Travel and Tourism Industry": [
            "Structure of the industry", "Types of tourism",
            "Travel organisations", "Sustainable tourism",
        ],
        "Features of Destinations": [
            "Destination appeal", "Tourist motivations and typologies",
            "Destination management",
        ],
        "Customer Service": [
            "Customer needs", "Quality service", "Handling complaints", "Selling skills",
        ],
        "Marketing in Travel and Tourism": [
            "The marketing mix", "Market research", "Promotion", "Branding",
        ],
        "Tourism Development": [
            "Economic impacts", "Social and cultural impacts",
            "Environmental impacts", "Managing tourism",
        ],
    },
    # ===================== O-LEVEL / IGCSE =====================
    "O_0452": {  # Accounting
        "Introduction to Accounting": [
            "The accounting equation", "Double entry bookkeeping",
            "Books of prime entry", "The ledger and trial balance",
        ],
        "Financial Statements": [
            "Income statement", "Statement of financial position",
            "Accruals and prepayments", "Depreciation",
        ],
        "Accounting for Different Entities": [
            "Sole traders", "Partnerships", "Limited companies",
            "Non-profit organisations",
        ],
        "Verification and Control": [
            "Bank reconciliation statements", "Control accounts",
            "Correction of errors", "Suspense accounts",
        ],
        "Analysis and Interpretation": [
            "Accounting ratios", "Interpreting accounts",
        ],
    },
    "O_0610": {  # Biology
        "Characteristics and Cells": [
            "Characteristics of living organisms", "Cell structure",
            "Movement in and out of cells", "Classification",
        ],
        "Molecules and Nutrition": [
            "Biological molecules", "Enzymes",
            "Nutrition in plants", "Nutrition in humans",
        ],
        "Transport and Respiration": [
            "Transport in plants", "Transport in animals",
            "Gas exchange", "Respiration",
        ],
        "Coordination and Homeostasis": [
            "The nervous system", "Hormones", "Homeostasis", "Excretion",
        ],
        "Reproduction and Inheritance": [
            "Reproduction in plants", "Reproduction in humans",
            "Inheritance", "Variation and selection",
        ],
        "Ecology and Human Impact": [
            "Organisms and their environment",
            "Human influences on ecosystems", "Biotechnology",
        ],
    },
    "O_0450": {  # Business Studies
        "Business Activity": [
            "Purpose and nature of business", "Classification of business",
            "Enterprise and entrepreneurs", "Business growth",
        ],
        "People in Business": [
            "Motivation", "Organisation and management",
            "Recruitment and training", "Communication",
        ],
        "Marketing": [
            "Market research", "The marketing mix", "Marketing strategy",
        ],
        "Operations Management": [
            "Production methods", "Costs and break-even",
            "Achieving quality", "Location decisions",
        ],
        "Finance": [
            "Sources of finance", "Cash flow forecasting",
            "Financial statements", "Ratio analysis",
        ],
    },
    "O_0620": {  # Chemistry
        "Matter and Atoms": [
            "Particulate nature of matter", "Atomic structure",
            "The periodic table", "Chemical bonding",
        ],
        "Stoichiometry": [
            "Formulae and equations", "The mole concept", "Reacting masses",
        ],
        "Electrochemistry and Energetics": [
            "Electrolysis", "Energy changes", "Rates of reaction",
            "Reversible reactions",
        ],
        "Acids, Bases and Salts": [
            "Acids and bases", "Preparation of salts", "Identification of ions",
        ],
        "Metals and Periodicity": [
            "Properties of metals", "The reactivity series", "Extraction of metals",
        ],
        "Organic and Environmental Chemistry": [
            "Fuels", "Alkanes and alkenes", "Alcohols", "Air and water",
        ],
    },
    "O_0478": {  # Computer Science
        "Data Representation": [
            "Binary systems", "Hexadecimal", "Data storage and compression",
        ],
        "Communication and Networking": [
            "Data transmission", "Networks", "The internet", "Security",
        ],
        "Hardware": [
            "Logic gates", "Input and output devices", "Storage devices", "The CPU",
        ],
        "Software": [
            "System software", "Application software", "Operating systems",
        ],
        "Programming and Databases": [
            "Algorithm design", "Pseudocode and flowcharts",
            "Programming concepts", "Databases",
        ],
    },
    "O_0500": {  # English Language
        "Reading Skills": [
            "Comprehension", "Summary writing",
            "Extended response to reading", "Analysing language",
        ],
        "Directed Writing": [
            "Writing for purpose", "Audience and tone",
            "Discursive writing", "Argumentative writing",
        ],
        "Composition": [
            "Descriptive writing", "Narrative writing", "Planning and structure",
        ],
        "Language Techniques": [
            "Vocabulary and style", "Grammar and accuracy", "Punctuation",
        ],
    },
    "O_0475": {  # English Literature
        "Poetry": [
            "Reading poetry", "Imagery and figurative language",
            "Form and structure", "Comparing poems",
        ],
        "Prose": [
            "Studying the novel", "Character and theme",
            "Setting and context", "Close reading",
        ],
        "Drama": [
            "Studying plays", "Dramatic devices",
            "Character and conflict", "Themes",
        ],
        "Critical Writing": [
            "Essay structure", "Using quotations", "Personal response",
        ],
    },
    "O_0680": {  # Environmental Management
        "The Lithosphere": [
            "Rocks and minerals", "Mining and extraction", "Soil and its management",
        ],
        "The Atmosphere": [
            "Atmosphere and climate", "Air pollution", "Climate change",
        ],
        "The Hydrosphere": [
            "Water resources", "Water pollution", "Managing water", "The oceans",
        ],
        "The Biosphere": [
            "Ecosystems", "Biodiversity", "Human impact", "Conservation",
        ],
        "Population and Sustainability": [
            "Population dynamics", "Resources and energy", "Sustainable development",
        ],
    },
    "O_0460": {  # Geography
        "Population and Settlement": [
            "Population dynamics", "Migration", "Settlement patterns", "Urbanisation",
        ],
        "The Natural Environment": [
            "Plate tectonics", "Rivers", "Coasts", "Weather and climate", "Ecosystems",
        ],
        "Economic Development": [
            "Agriculture", "Industry", "Tourism", "Energy", "Development",
        ],
        "Geographical Skills": [
            "Map skills", "Data collection", "Interpreting data",
        ],
    },
    "O_0470": {  # History
        "The Nineteenth Century": [
            "International relations", "Nationalism", "Empire and colonialism",
        ],
        "The First World War": [
            "Causes of the First World War", "The course of the war", "Peace treaties",
        ],
        "Between the Wars": [
            "The League of Nations", "Rise of dictatorships",
            "Causes of the Second World War",
        ],
        "The Cold War": [
            "Origins of the Cold War", "Crises and conflicts", "End of the Cold War",
        ],
        "Depth Study": [
            "Germany 1918-1945", "Russia and the USSR", "The United States",
        ],
    },
    "O_0625": {  # Physics
        "General Physics": [
            "Measurements", "Motion", "Mass, weight and density",
            "Forces", "Energy, work and power", "Pressure",
        ],
        "Thermal Physics": [
            "Kinetic theory", "Thermal properties", "Transfer of thermal energy",
        ],
        "Waves": [
            "General wave properties", "Light", "Sound",
            "The electromagnetic spectrum",
        ],
        "Electricity and Magnetism": [
            "Magnetism", "Static electricity", "Current electricity",
            "Electrical circuits", "Electromagnetic effects",
        ],
        "Atomic Physics": [
            "The nuclear atom", "Radioactivity",
        ],
    },
}


def _slug(text: str) -> str:
    keep = [c.lower() if c.isalnum() else "_" for c in text]
    out = "".join(keep)
    while "__" in out:
        out = out.replace("__", "_")
    return out.strip("_")[:40]


def build_tree(subject_content: dict) -> list:
    """Convert {topic: [lessons]} into the app's tree node shape."""
    tree = []
    for t_index, (topic_title, lessons) in enumerate(subject_content.items(), start=1):
        topic_id = f"t{t_index}_{_slug(topic_title)}"
        children = []
        for l_index, lesson_title in enumerate(lessons, start=1):
            children.append({
                "id":    f"{topic_id}_l{l_index}",
                "title": lesson_title,
                "type":  "lesson",
            })
        tree.append({
            "id":       topic_id,
            "title":    topic_title,
            "type":     "topic",
            "children": children,
        })
    return tree


def main():
    dry_run = "--dry-run" in sys.argv
    structures = {rid: {"tree": build_tree(content)} for rid, content in SYLLABI.items()}

    total_topics = sum(len(s["tree"]) for s in structures.values())
    total_lessons = sum(
        len(node["children"]) for s in structures.values() for node in s["tree"]
    )
    print(f"Prepared {len(structures)} subjects, "
          f"{total_topics} topics, {total_lessons} lessons.")

    if dry_run:
        print(json.dumps(structures["A_9709"], indent=2))
        print("\n[dry-run] Nothing sent.")
        return

    url = f"{DATA_API_BASE_URL}/admin/seed-structures"
    print(f"POST {url}")
    resp = requests.post(url, json={"structures": structures}, timeout=120)
    print(f"HTTP {resp.status_code}")
    try:
        results = resp.json().get("results", {})
        ok = sum(1 for r in results.values() if r.get("ok"))
        print(f"Seeded OK: {ok}/{len(results)}")
        failures = {k: v for k, v in results.items() if not v.get("ok")}
        if failures:
            print("Failures:", json.dumps(failures, indent=2))
    except Exception:
        print(resp.text[:1000])


if __name__ == "__main__":
    main()