import json
import pathlib

TASK_CATEGORIES: dict[str, list[str]] = {
    "histone_marks": [
        "h2afz", "h3k27ac", "h3k27me3", "h3k36me3",
        "h3k4me1", "h3k4me2", "h3k4me3", "h3k9ac", "h3k9me3", "h4k20me1",
    ],
    "promoter": ["promoter_all", "promoter_tata", "promoter_no_tata"],
    "enhancer": ["enhancers", "enhancers_types"],
    "splice_site": ["splice_sites_all", "splice_sites_donors", "splice_sites_acceptors"],
    "variant_effect": [
        "variant_effect_pathogenic_clinvar", "variant_effect_pathogenic_omim", "variant_effect_causal_eqtl",
    ],
    "rna_expression": ["bulk_rna_expression"],
}

# Inverse map: task_name -> category label
TASK_TO_CATEGORY: dict[str, str] = {
    task: cat for cat, tasks in TASK_CATEGORIES.items() for task in tasks
}

# Mean sequence lengths in bp per task, populated by tools/compute_seq_lengths.py
_SEQ_LEN_PATH = pathlib.Path(__file__).parent.parent / "tools" / "task_seq_lengths.json"
TASK_SEQ_LENGTHS: dict[str, int] = (
    json.loads(_SEQ_LEN_PATH.read_text()) if _SEQ_LEN_PATH.exists() else {}
)

CLASSIFICATION_METRICS: dict[str, str] = {
    "mcc_test": "MCC",
    "accuracy_test": "Accuracy",
    "weighted_f1_test": "Weighted F1",
    "macro_f1_test": "Macro F1",
    "auprc_test": "AUPRC",
}

CATEGORY_DISPLAY_NAMES: dict[str, str] = {
    "histone_marks": "Histone Marks",
    "promoter": "Promoter",
    "enhancer": "Enhancer",
    "splice_site": "Splice Site",
    "variant_effect": "Variant Effect",
    "rna_expression": "RNA Expression",
    "other": "Other",
}

# One color per task category — Loka-brand-anchored categorical palette, validated
# for CVD-safe adjacency (see dataviz skill / src/plots.py LOKA_PALETTE). Fixed
# hue order, never cycled — distinct from _MODEL_PALETTE ordering in plots.py.
CATEGORY_COLORS: dict[str, str] = {
    "histone_marks":  "#1877F2",  # blue   (Loka brand)
    "promoter":       "#EB6834",  # orange
    "enhancer":       "#0AA88F",  # teal   (Loka brand green, re-stepped for chroma/lightness)
    "splice_site":    "#D9A61F",  # yellow (Loka brand, darkened for contrast)
    "variant_effect": "#E87BA4",  # magenta
    "rna_expression": "#008300",  # green
    "other":          "#4A3AA7",  # violet
}