backend/knowledge_base.py

# -*- coding: utf-8 -*-
"""
CropAI Agronomic Knowledge Base - Master Integration Hub
Contains curated pathology details, symptoms, exact CIBRC chemical prescriptions (calibrated per 16L/20L knapsack tank),
organic bio-controls, seasonal triggers, and historical outbreak contexts for 127+ crop diseases, pests,
insects, and physiological disorders across Chhattisgarh (Surguja, Bilaspur, Raipur, Durg, Bastar, etc.)
and Pan-India field, pulse, oilseed, cash, vegetable, and fruit crops.
"""

from typing import Dict, Any, Optional

# Import domain modules
from .data_cereals import CEREALS_KNOWLEDGE_BASE
from .data_pulses import PULSES_KNOWLEDGE_BASE
from .data_oilseeds import OILSEEDS_KNOWLEDGE_BASE
from .data_cashcrops import CASH_CROPS_KNOWLEDGE_BASE
from .data_vegetables import VEGETABLES_KNOWLEDGE_BASE
from .data_fruits import FRUITS_KNOWLEDGE_BASE
from .data_benchmark import BENCHMARK_KNOWLEDGE_BASE
from .hf_dataset_sync import get_all_hf_aliases

# Assemble unified agronomic knowledge base
DISEASE_KNOWLEDGE_BASE: Dict[str, Dict[str, Any]] = {}
DISEASE_KNOWLEDGE_BASE.update(CEREALS_KNOWLEDGE_BASE)
DISEASE_KNOWLEDGE_BASE.update(PULSES_KNOWLEDGE_BASE)
DISEASE_KNOWLEDGE_BASE.update(OILSEEDS_KNOWLEDGE_BASE)
DISEASE_KNOWLEDGE_BASE.update(CASH_CROPS_KNOWLEDGE_BASE)
DISEASE_KNOWLEDGE_BASE.update(VEGETABLES_KNOWLEDGE_BASE)
DISEASE_KNOWLEDGE_BASE.update(FRUITS_KNOWLEDGE_BASE)
DISEASE_KNOWLEDGE_BASE.update(BENCHMARK_KNOWLEDGE_BASE)


# =========================================================================
# Hugging Face / Swin / MobileNet / ResNet / YOLO Label Aliases
# =========================================================================
HF_CLASS_MAP: Dict[str, str] = {
    # Rice / Paddy Models
    'Bacterialblight': 'Rice___Bacterial_Leaf_Blight',
    'Bacterial Leaf Blight': 'Rice___Bacterial_Leaf_Blight',
    'Blast': 'Rice___Blast',
    'Leaf Blast': 'Rice___Blast',
    'Neck Blast': 'Rice___Blast',
    'Brownspot': 'Rice___Brown_Spot',
    'Brown Spot': 'Rice___Brown_Spot',
    'Healthy': 'Rice___healthy',
    'Healthy Rice Leaf': 'Rice___healthy',
    'Rice healthy': 'Rice___healthy',
    'Tungro': 'Rice___Tungro',
    'Leaf Scald': 'Rice___Sheath_Blight',
    'Leaf scald': 'Rice___Sheath_Blight',
    'Sheath Blight': 'Rice___Sheath_Blight',
    'False Smut': 'Rice___False_Smut',
    'Khaira Disease': 'Rice___Khaira_Disease',
    'Yellow Stem Borer': 'Rice___Yellow_Stem_Borer',
    'Gall Midge': 'Rice___Gall_Midge',
    'Brown Plant Hopper': 'Rice___Brown_Plant_Hopper',
    'BPH': 'Rice___Brown_Plant_Hopper',
    'Leaf Folder': 'Rice___Leaf_Folder',
    'Rice Hispa': 'Rice___Hispa',
    'Gundhi Bug': 'Rice___Gundhi_Bug',
    'Bakanae': 'Rice___Bakanae',

    # Wheat
    'Wheat Leaf Rust': 'Wheat___Brown_Rust',
    'Wheat Brown Rust': 'Wheat___Brown_Rust',
    'Wheat Yellow Rust': 'Wheat___Yellow_Rust',
    'Wheat Stripe Rust': 'Wheat___Yellow_Rust',
    'Wheat Black Rust': 'Wheat___Black_Rust',
    'Wheat Stem Rust': 'Wheat___Black_Rust',
    'Wheat Loose Smut': 'Wheat___Loose_Smut',
    'Wheat Karnal Bunt': 'Wheat___Karnal_Bunt',
    'Wheat Powdery Mildew': 'Wheat___Powdery_Mildew',
    'Wheat Termite': 'Wheat___Termites',
    'Wheat healthy': 'Wheat___healthy',

    # Maize / Corn
    'Corn (Maize) with Cercospora and Gray Leaf Spot': 'Corn_(maize)___Cercospora_leaf_spot Gray_leaf_spot',
    'Corn (Maize) with Common Rust': 'Corn_(maize)___Common_rust_',
    'Corn (Maize) with Northern Leaf Blight': 'Corn_(maize)___Northern_Leaf_Blight',
    'Healthy Corn (Maize) Plant': 'Corn_(maize)___healthy',
    'Corn Fall Armyworm': 'Maize___Fall_Armyworm',
    'Fall Armyworm': 'Maize___Fall_Armyworm',
    'Corn Stem Borer': 'Maize___Stem_Borer',

    # Pulses (Chickpea, Pigeonpea, Urad, Moong, Tiwda)
    'Chickpea Pod Borer': 'Chickpea___Pod_Borer',
    'Gram Pod Borer': 'Chickpea___Pod_Borer',
    'Chickpea Fusarium Wilt': 'Chickpea___Fusarium_Wilt',
    'Chickpea Ascochyta Blight': 'Chickpea___Ascochyta_Blight',
    'Chickpea Dry Root Rot': 'Chickpea___Dry_Root_Rot',
    'Chickpea Collar Rot': 'Chickpea___Collar_Rot',
    'Chickpea healthy': 'Chickpea___healthy',
    'Pigeonpea Wilt': 'Pigeonpea___Fusarium_Wilt',
    'Arhar Wilt': 'Pigeonpea___Fusarium_Wilt',
    'Pigeonpea Sterility Mosaic': 'Pigeonpea___Sterility_Mosaic_Disease',
    'Pigeonpea Pod Borer': 'Pigeonpea___Pod_Borer',
    'Pigeonpea Pod Fly': 'Pigeonpea___Pod_Fly',
    'Pigeonpea Phytophthora Blight': 'Pigeonpea___Phytophthora_Blight',
    'Pigeonpea healthy': 'Pigeonpea___healthy',
    'Urad Yellow Mosaic': 'Urad_Moong___Yellow_Mosaic_Virus',
    'Moong Yellow Mosaic': 'Urad_Moong___Yellow_Mosaic_Virus',
    'Urad Cercospora Leaf Spot': 'Urad_Moong___Cercospora_Leaf_Spot',
    'Urad Powdery Mildew': 'Urad_Moong___Powdery_Mildew',
    'Urad healthy': 'Urad_Moong___healthy',
    'Tiwda Downy Mildew': 'Lathyrus___Downy_Mildew',
    'Tiwda Rust': 'Lathyrus___Rust',
    'Tiwda healthy': 'Lathyrus___healthy',

    # Oilseeds (Soybean, Mustard, Groundnut, Niger)
    'Soybean Girdle Beetle': 'Soybean___Girdle_Beetle',
    'Soybean Yellow Mosaic': 'Soybean___Yellow_Mosaic_Virus',
    'Soybean Rust': 'Soybean___Rust',
    'Soybean Anthracnose': 'Soybean___Anthracnose',
    'Soybean Stem Fly': 'Soybean___Stem_Fly',
    'Soybean Defoliator': 'Soybean___Spodoptera_Litura',
    'Healthy Soybean Plant': 'Soybean___healthy',
    'Soybean healthy': 'Soybean___healthy',
    'Mustard Aphid': 'Mustard___Aphid',
    'Mustard White Rust': 'Mustard___White_Rust',
    'Mustard Alternaria Blight': 'Mustard___Alternaria_Blight',
    'Mustard Powdery Mildew': 'Mustard___Powdery_Mildew',
    'Mustard Sclerotinia Stem Rot': 'Mustard___Sclerotinia_Stem_Rot',
    'Mustard healthy': 'Mustard___healthy',
    'Groundnut Tikka Leaf Spot': 'Groundnut___Tikka_Leaf_Spot',
    'Groundnut Collar Rot': 'Groundnut___Collar_Rot',
    'Groundnut Rust': 'Groundnut___Rust',
    'Groundnut healthy': 'Groundnut___healthy',

    # Cash Crops (Cotton, Sugarcane)
    'Cotton Pink Bollworm': 'Cotton___Pink_Bollworm',
    'Cotton Whitefly': 'Cotton___Whitefly',
    'Cotton Bacterial Blight': 'Cotton___Bacterial_Blight',
    'Cotton Leaf Curl Virus': 'Cotton___Leaf_Curl_Virus',
    'Cotton Boll Rot': 'Cotton___Boll_Rot_Complex',
    'Cotton healthy': 'Cotton___healthy',
    'Sugarcane Red Rot': 'Sugarcane___Red_Rot',
    'Sugarcane Early Shoot Borer': 'Sugarcane___Early_Shoot_Borer',
    'Sugarcane Pyrilla': 'Sugarcane___Pyrilla',
    'Sugarcane Smut': 'Sugarcane___Smut',
    'Sugarcane healthy': 'Sugarcane___healthy',

    # Vegetables (Tomato, Potato, Chilli, Brinjal, Okra)
    'Tomato with Bacterial Spot': 'Tomato___Bacterial_spot',
    'Tomato_Bacterial_spot': 'Tomato___Bacterial_spot',
    'Tomato Bacterial Spot': 'Tomato___Bacterial_spot',
    'Tomato with Early Blight': 'Tomato___Early_blight',
    'Tomato_Early_blight': 'Tomato___Early_blight',
    'Tomato Early Blight': 'Tomato___Early_blight',
    'Tomato with Late Blight': 'Tomato___Late_blight',
    'Tomato_Late_blight': 'Tomato___Late_blight',
    'Tomato Late Blight': 'Tomato___Late_blight',
    'Tomato with Leaf Mold': 'Tomato___Leaf_Mold',
    'Tomato_Leaf_Mold': 'Tomato___Leaf_Mold',
    'Tomato with Septoria Leaf Spot': 'Tomato___Septoria_leaf_spot',
    'Tomato_Septoria_leaf_spot': 'Tomato___Septoria_leaf_spot',
    'Tomato with Spider Mites or Two-spotted Spider Mite': 'Tomato___Spider_mites Two-spotted_spider_mite',
    'Tomato_Spider_mites_Two_spotted_spider_mite': 'Tomato___Spider_mites Two-spotted_spider_mite',
    'Tomato with Target Spot': 'Tomato___Target_Spot',
    'Tomato__Target_Spot': 'Tomato___Target_Spot',
    'Tomato Yellow Leaf Curl Virus': 'Tomato___Tomato_Yellow_Leaf_Curl_Virus',
    'Tomato__Tomato_YellowLeaf__Curl_Virus': 'Tomato___Tomato_Yellow_Leaf_Curl_Virus',
    'Tomato Mosaic Virus': 'Tomato___Tomato_mosaic_virus',
    'Tomato__Tomato_mosaic_virus': 'Tomato___Tomato_mosaic_virus',
    'Tomato Fruit Borer': 'Tomato___Fruit_Borer',
    'Healthy Tomato Plant': 'Tomato___healthy',
    'Tomato_healthy': 'Tomato___healthy',
    'Tomato healthy': 'Tomato___healthy',

    'Potato with Early Blight': 'Potato___Early_blight',
    'Potato_Early_blight': 'Potato___Early_blight',
    'Potato Early Blight': 'Potato___Early_blight',
    'Potato with Late Blight': 'Potato___Late_blight',
    'Potato_Late_blight': 'Potato___Late_blight',
    'Potato Late Blight': 'Potato___Late_blight',
    'Potato Bacterial Wilt': 'Potato___Bacterial_Wilt',
    'Healthy Potato Plant': 'Potato___healthy',
    'Potato_healthy': 'Potato___healthy',
    'Potato healthy': 'Potato___healthy',

    'Chilli Leaf Curl Virus': 'Chilli___Leaf_Curl_Virus',
    'Chilli Anthracnose Fruit Rot': 'Chilli___Anthracnose_Dieback',
    'Chilli Thrips': 'Chilli___Thrips',
    'Chilli healthy': 'Chilli___healthy',

    'Brinjal Shoot and Fruit Borer': 'Brinjal___Shoot_and_Fruit_Borer',
    'Brinjal Phomopsis Blight': 'Brinjal___Phomopsis_Blight_Fruit_Rot',
    'Brinjal Little Leaf': 'Brinjal___Little_Leaf_Disease',
    'Brinjal healthy': 'Brinjal___healthy',

    'Okra Yellow Vein Mosaic Virus': 'Okra___Yellow_Vein_Mosaic_Virus',
    'Okra Shoot and Fruit Borer': 'Okra___Shoot_and_Fruit_Borer',
    'Okra Powdery Mildew': 'Okra___Powdery_Mildew',
    'Okra healthy': 'Okra___healthy',

    # Fruits & Alliums (Mango, Papaya, Citrus, Onion, Garlic, Cucurbits)
    'Mango Powdery Mildew': 'Mango___Powdery_Mildew',
    'Mango Anthracnose': 'Mango___Anthracnose',
    'Mango Hopper': 'Mango___Hopper',
    'Mango Malformation': 'Mango___Malformation',
    'Papaya Ringspot Virus': 'Papaya___Ringspot_Virus',
    'Papaya Mealybug': 'Papaya___Mealybug',
    'Citrus Canker': 'Citrus___Canker',
    'Citrus Greening': 'Citrus___Greening_Huanglongbing',
    'Citrus Leaf Miner': 'Citrus___Leaf_Miner',
    'Orange with Citrus Greening': 'Orange___Haunglongbing_(Citrus_greening)',
    'Onion Purple Blotch': 'Onion_Garlic___Purple_Blotch',
    'Onion Thrips': 'Onion_Garlic___Thrips',
    'Garlic Stemphylium Blight': 'Onion_Garlic___Stemphylium_Blight',
    'Cucurbit Downy Mildew': 'Cucurbits___Downy_Mildew',
    'Cucurbit Powdery Mildew': 'Cucurbits___Powdery_Mildew',
    'Cucurbit Fruit Fly': 'Cucurbits___Fruit_Fly',

    # Benchmark PlantVillage
    'Apple Scab': 'Apple___Apple_scab',
    'Apple with Black Rot': 'Apple___Black_rot',
    'Cedar Apple Rust': 'Apple___Cedar_apple_rust',
    'Healthy Apple': 'Apple___healthy',
    'Healthy Blueberry Plant': 'Blueberry___healthy',
    'Cherry with Powdery Mildew': 'Cherry___Powdery_mildew',
    'Healthy Cherry Plant': 'Cherry___healthy',
    'Grape with Black Rot': 'Grape___Black_rot',
    'Grape with Esca (Black Measles)': 'Grape___Esca_(Black_Measles)',
    'Grape with Isariopsis Leaf Spot': 'Grape___Leaf_blight_(Isariopsis_Leaf_Spot)',
    'Healthy Grape Plant': 'Grape___healthy',
    'Peach with Bacterial Spot': 'Peach___Bacterial_spot',
    'Healthy Peach Plant': 'Peach___healthy',
    'Bell Pepper with Bacterial Spot': 'Pepper,_bell___Bacterial_spot',
    'Pepper__bell___Bacterial_spot': 'Pepper,_bell___Bacterial_spot',
    'Healthy Bell Pepper Plant': 'Pepper,_bell___healthy',
    'Pepper__bell___healthy': 'Pepper,_bell___healthy',
    'Healthy Raspberry Plant': 'Raspberry___healthy',
    'Squash with Powdery Mildew': 'Squash___Powdery_mildew',
    'Strawberry with Leaf Scorch': 'Strawberry___Leaf_scorch',
    'Healthy Strawberry Plant': 'Strawberry___healthy',
}
HF_CLASS_MAP.update(get_all_hf_aliases())

DEFAULT_UNKNOWN_DISEASE: Dict[str, Any] = {
    "name": "General Foliar Condition / Mild Anomaly",
    "scientific_name": "Pathology Under Observation",
    "crop": "General Crop",
    "category": "Anomaly",
    "severity": "Moderate",
    "is_healthy": False,
    "season_and_weather": "Variable conditions; monitor field microclimate closely",
    "economic_injury_level": "Inspect 20 random plants across field diagonals",
    "symptoms": [
        "Visual anomaly detected on the leaf surface",
        "Possible localized chlorosis, pigmentation irregularity, or mechanical injury",
        "Requires further inspection in bright natural daylight"
    ],
    "treatment": {
        "immediate_action": [
            "Isolate the affected branch or plant to prevent potential contagion",
            "Ensure leaf is clean and retake photo in bright natural daylight"
        ],
        "organic_control": [
            "Spray neem oil solution (0.5% - 80ml per 16L tank) as a broad-spectrum preventive bio-agent",
            "Ensure balanced compost and micro-nutrient foliar spray"
        ],
        "chemical_control": [
            "Consult with an agricultural extension officer (KVK / Krishi Mitr) before applying systemic chemical sprays"
        ]
    },
    "prevention": [
        "Maintain good farm sanitation and remove weed hosts",
        "Monitor crop weekly and avoid prolonged leaf wetness"
    ],
    "expert_consultation_recommended": True
}


def get_disease_info(class_id: str) -> Dict[str, Any]:
    """
    Look up disease metadata from canonical class ID, HuggingFace label, or normalized query.
    Returns comprehensive agronomic details including tank dosages, organic controls, and symptoms.
    """
    # 1. Direct dictionary match
    if class_id in DISEASE_KNOWLEDGE_BASE:
        info = DISEASE_KNOWLEDGE_BASE[class_id].copy()
        info["id"] = class_id
        return info

    # 2. Check Hugging Face explicit alias map
    if class_id in HF_CLASS_MAP:
        mapped_key = HF_CLASS_MAP[class_id]
        if mapped_key in DISEASE_KNOWLEDGE_BASE:
            info = DISEASE_KNOWLEDGE_BASE[mapped_key].copy()
            info["id"] = mapped_key
            return info

    # 3. Normalized fuzzy search
    clean_id = class_id.replace(" ", "_").replace("-", "_").replace("__", "_").lower()
    for key, data in DISEASE_KNOWLEDGE_BASE.items():
        clean_key = key.replace(" ", "_").replace("-", "_").replace("__", "_").lower()
        if clean_key == clean_id:
            info = data.copy()
            info["id"] = key
            return info

    # 4. Partial substring match in knowledge base keys
    for key, data in DISEASE_KNOWLEDGE_BASE.items():
        if class_id.lower() in key.lower() or key.lower() in class_id.lower():
            info = data.copy()
            info["id"] = key
            return info

    # 5. Fallback representation
    info = DEFAULT_UNKNOWN_DISEASE.copy()
    parts = class_id.split("___")
    if len(parts) == 2:
        info["crop"] = parts[0].replace("_", " ").title()
        info["name"] = f"{info['crop']} {parts[1].replace('_', ' ').title()}"
    else:
        info["crop"] = "Unknown Crop"
        info["name"] = class_id.replace("_", " ").title()
    info["id"] = class_id
    return info