test_crossref_sorting.py

#!/usr/bin/env python3
"""
测试Crossref API的排序机制
验证排序是否真的按引用量进行
"""

import requests
import json

def test_crossref_sorting():
    """测试Crossref API的排序机制"""
    url = "https://api.crossref.org/works"
    
    headers = {
        'User-Agent': 'Academic-Reviewer-System/1.0 (mailto:[email protected])'
    }
    
    # 测试查询
    query = "machine learning"
    
    print("=== 测试Crossref API排序机制 ===")
    print(f"查询: {query}")
    print("=" * 60)
    
    # 测试不同的排序方式
    sort_options = ["published", "relevance", "deposited"]
    
    for sort_option in sort_options:
        print(f"\n--- 排序方式: {sort_option} ---")
        
        params = {
            "query": query,
            "rows": 10,
            "sort": sort_option,
            "order": "desc",
            "select": "DOI,title,author,container-title,published-print,published-online,is-referenced-by-count"
        }
        
        try:
            response = requests.get(url, params=params, headers=headers, timeout=30)
            response.raise_for_status()
            data = response.json()
            
            items = data.get("message", {}).get("items", [])
            
            print(f"返回 {len(items)} 个结果")
            
            # 显示原始排序的引用量
            print("原始排序的引用量:")
            original_citations = []
            for i, item in enumerate(items, 1):
                title = item.get('title', ['N/A'])[0] if item.get('title') else 'N/A'
                if len(title) > 40:
                    title = title[:40] + "..."
                cited_count = item.get('is-referenced-by-count', 0)
                original_citations.append(cited_count)
                print(f"  {i}. {title} (引用: {cited_count})")
            
            # 按引用量重新排序
            sorted_items = sorted(items, key=lambda x: x.get('is-referenced-by-count', 0), reverse=True)
            
            print("\n按引用量重新排序:")
            sorted_citations = []
            for i, item in enumerate(sorted_items, 1):
                title = item.get('title', ['N/A'])[0] if item.get('title') else 'N/A'
                if len(title) > 40:
                    title = title[:40] + "..."
                cited_count = item.get('is-referenced-by-count', 0)
                sorted_citations.append(cited_count)
                print(f"  {i}. {title} (引用: {cited_count})")
            
            # 比较排序差异
            print(f"\n排序差异分析:")
            print(f"原始排序引用量: {original_citations}")
            print(f"按引用量排序: {sorted_citations}")
            
            # 检查是否相同
            if original_citations == sorted_citations:
                print("✅ 原始排序与按引用量排序相同")
            else:
                print("❌ 原始排序与按引用量排序不同")
                print("   说明Crossref API没有按引用量排序")
            
            # 统计引用量分布
            non_zero_citations = [c for c in original_citations if c > 0]
            print(f"非零引用量数量: {len(non_zero_citations)}/{len(original_citations)}")
            if non_zero_citations:
                print(f"平均引用量: {sum(non_zero_citations) / len(non_zero_citations):.2f}")
                print(f"最大引用量: {max(non_zero_citations)}")
            
        except Exception as e:
            print(f"错误: {str(e)}")
        
        print("-" * 50)

def test_high_citation_query():
    """测试高引用量查询"""
    url = "https://api.crossref.org/works"
    
    headers = {
        'User-Agent': 'Academic-Reviewer-System/1.0 (mailto:[email protected])'
    }
    
    # 使用更可能产生高引用量的查询
    high_citation_queries = [
        "CRISPR",
        "machine learning",
        "deep learning",
        "artificial intelligence"
    ]
    
    print("\n=== 测试高引用量查询 ===")
    print("=" * 60)
    
    for query in high_citation_queries:
        print(f"\n--- 查询: {query} ---")
        
        params = {
            "query": query,
            "rows": 5,
            "sort": "relevance",  # 使用相关性排序
            "order": "desc",
            "select": "DOI,title,author,container-title,published-print,published-online,is-referenced-by-count"
        }
        
        try:
            response = requests.get(url, params=params, headers=headers, timeout=30)
            response.raise_for_status()
            data = response.json()
            
            items = data.get("message", {}).get("items", [])
            
            # 按引用量排序
            sorted_items = sorted(items, key=lambda x: x.get('is-referenced-by-count', 0), reverse=True)
            
            print(f"按引用量排序的前3个结果:")
            for i, item in enumerate(sorted_items[:3], 1):
                title = item.get('title', ['N/A'])[0] if item.get('title') else 'N/A'
                if len(title) > 50:
                    title = title[:50] + "..."
                cited_count = item.get('is-referenced-by-count', 0)
                
                # 获取发表年份
                pub_date = item.get('published-print', {}) or item.get('published-online', {})
                pub_year = pub_date.get('date-parts', [[None]])[0][0] if pub_date else 'N/A'
                
                print(f"  {i}. {title}")
                print(f"     引用量: {cited_count}, 年份: {pub_year}")
            
            # 统计
            citations = [item.get('is-referenced-by-count', 0) for item in items]
            non_zero = [c for c in citations if c > 0]
            print(f"  总结果: {len(items)}, 有引用量: {len(non_zero)}, 平均引用量: {sum(non_zero) / len(non_zero) if non_zero else 0:.2f}")
            
        except Exception as e:
            print(f"错误: {str(e)}")

if __name__ == "__main__":
    test_crossref_sorting()
    test_high_citation_query()