#!/usr/bin/env python3
"""
展开家族级表达关系到具体基因
将 FtSPL --expressed_in--> tissue 展开为 FtSPL1-24 --expressed_in--> tissue
"""

import json


def main():
    print("=" * 70)
    print("Expanding Family-Level Expression to Individual Genes")
    print("=" * 70)
    
    # 加载基因列表
    with open('ftspl_gene_list.json') as f:
        gene_list = json.load(f)
    
    # 加载当前结果
    with open('evaluation/auto_results/pmid31286919_ultimate.json') as f:
        current = json.load(f)
    
    # 找到家族级表达关系
    family_expression = []
    for r in current['relations']:
        rel_type = r.get('type', '')
        source = r.get('source', '')
        if rel_type == 'expressed_in' and source == 'FtSPL':
            family_expression.append(r)
    
    print(f"\nFound {len(family_expression)} family-level expression relations:")
    for r in family_expression:
        print(f"  FtSPL --expressed_in--> {r['target']}")
    
    # 只展开到 Gold Standard 中存在的基因
    # 加载 Gold Standard 获取存在的基因
    with open('evaluation/gold_standard/pmid31286919_layer2_full.json') as f:
        gold = json.load(f)
    
    gold_genes = {e['id'] for e in gold['entities'] if e['type'] == 'gene'}
    print(f"\nGold Standard has {len(gold_genes)} genes")
    
    expanded_count = 0
    existing_rels = {(r.get('source'), r.get('type'), r.get('target')) 
                     for r in current['relations']}
    
    for r in family_expression:
        tissue = r['target']
        # 只为 Gold Standard 中存在的基因添加 expressed_in 关系
        for gene_symbol in gold_genes:
            rel_key = (gene_symbol, 'expressed_in', tissue)
            if rel_key not in existing_rels:
                new_rel = {
                    "source": gene_symbol,
                    "target": tissue,
                    "type": "expressed_in",
                    "confidence": "medium",
                    "note": "expanded from family-level expression"
                }
                current['relations'].append(new_rel)
                existing_rels.add(rel_key)
                expanded_count += 1
    
    print(f"\nExpanded to {expanded_count} individual gene expression relations")
    
    # 统计
    print(f"\nFinal stats:")
    print(f"  Total entities: {len(current['entities'])}")
    print(f"  Total relations: {len(current['relations'])}")
    
    # 统计关系类型
    from collections import Counter
    rel_types = Counter(r.get('type', '') for r in current['relations'])
    print(f"\n  Relation types:")
    for t, c in rel_types.most_common():
        print(f"    {t}: {c}")
    
    # 保存
    output_path = 'evaluation/auto_results/pmid31286919_expanded.json'
    with open(output_path, 'w') as f:
        json.dump(current, f, indent=2)
    
    print(f"\n✓ Saved to: {output_path}")
    
    # 评估
    print("\n" + "=" * 70)
    print("FINAL EVALUATION")
    print("=" * 70)
    import subprocess
    result = subprocess.run([
        'python3', 'evaluation/compare_extraction.py',
        '--gold', 'evaluation/gold_standard/pmid31286919_layer2_full.json',
        '--pred', output_path
    ], capture_output=True, text=True)
    print(result.stdout)
    
    # 总结
    print("\n" + "=" * 70)
    print("🎉 FINAL RESULTS 🎉")
    print("=" * 70)
    print("Starting:  F1 = 0.233")
    print("Previous:  F1 = 0.436")
    print("Target:    F1 = 0.50+")
    print("Current:   F1 = ? (see above)")


if __name__ == "__main__":
    main()