arxiv-pulse 0.5.0__py3-none-any.whl → 0.6.1__py3-none-any.whl

arxiv_pulse/cli.py

@@ -10,14 +10,15 @@ from pathlib import Path
 import click
 from dotenv import load_dotenv
 import json
-from datetime import datetime, timedelta
+from datetime import datetime, timedelta, timezone
 import questionary
+import wcwidth
 
 from arxiv_pulse.config import Config
 from arxiv_pulse.arxiv_crawler import ArXivCrawler
 from arxiv_pulse.summarizer import PaperSummarizer
 from arxiv_pulse.report_generator import ReportGenerator
-from arxiv_pulse.output_manager import output
+from arxiv_pulse.output_manager import output, OutputLevel
 from arxiv_pulse.search_engine import SearchEngine, SearchFilter
 from arxiv_pulse.__version__ import __version__
 
@@ -234,7 +235,6 @@ def setup_environment(directory: Path):
         # 创建必要的目录
         os.makedirs("data", exist_ok=True)
         os.makedirs("reports", exist_ok=True)
-        os.makedirs("logs", exist_ok=True)
 
         # 加载 .env 文件（如果存在）
         env_file = directory / ".env"
@@ -267,12 +267,14 @@ def setup_environment(directory: Path):
         Config.AI_MODEL = os.getenv("AI_MODEL", "DeepSeek-V3.2-Thinking")
         Config.AI_BASE_URL = os.getenv("AI_BASE_URL", "https://llmapi.paratera.com")
         Config.SUMMARY_MAX_TOKENS = int(os.getenv("SUMMARY_MAX_TOKENS", "2000"))
-        Config.SUMMARY_SENTENCES_LIMIT = int(os.getenv("SUMMARY_SENTENCES_LIMIT", "3"))
         Config.TOKEN_PRICE_PER_MILLION = float(os.getenv("TOKEN_PRICE_PER_MILLION", "3.0"))
-        Config.MAX_RESULTS_INITIAL = int(os.getenv("MAX_RESULTS_INITIAL", "100"))
-        Config.MAX_RESULTS_DAILY = int(os.getenv("MAX_RESULTS_DAILY", "20"))
+        Config.MAX_RESULTS_INITIAL = int(os.getenv("MAX_RESULTS_INITIAL", "10000"))
+        Config.MAX_RESULTS_DAILY = int(os.getenv("MAX_RESULTS_DAILY", "500"))
         Config.YEARS_BACK = int(os.getenv("YEARS_BACK", "3"))
-        Config.IMPORTANT_PAPERS_FILE = os.getenv("IMPORTANT_PAPERS_FILE", "important_papers.txt")
+        Config.IMPORTANT_PAPERS_FILE = os.getenv("IMPORTANT_PAPERS_FILE", "data/important_papers.txt")
+        Config.ARXIV_MAX_RESULTS = int(os.getenv("ARXIV_MAX_RESULTS", "30000"))
+        Config.ARXIV_SORT_BY = os.getenv("ARXIV_SORT_BY", "submittedDate")
+        Config.ARXIV_SORT_ORDER = os.getenv("ARXIV_SORT_ORDER", "descending")
         Config.REPORT_MAX_PAPERS = int(os.getenv("REPORT_MAX_PAPERS", "50"))
 
         # 更新 SEARCH_QUERIES
@@ -407,54 +409,74 @@ def print_banner_custom(fields):
     else:
         field_str = f"{fields[0]} • {fields[1]} • {fields[2]} • {fields[3]}"
 
-    # 计算居中位置 (横幅宽度为55字符，边框各占1字符，内容宽度53字符)
-    # 第一行标题："arXiv Pulse - 文献追踪系统" (25字符)
-    # 需要将字段字符串居中显示
+    # 横幅尺寸
     banner_width = 55
     content_width = 53
 
-    # 创建横幅
+    # 辅助函数：计算字符串显示宽度
+    def display_width(text):
+        return wcwidth.wcswidth(text)
+
+    # 辅助函数：截断字符串到指定显示宽度，添加省略号
+    def truncate_to_width(text, max_width):
+        if display_width(text) <= max_width:
+            return text
+        # 逐步减少字符直到宽度合适
+        result = ""
+        for char in text:
+            if display_width(result + char) > max_width - 3:  # 为"..."留出空间
+                break
+            result += char
+        return result + "..." if result else "..."  # 至少返回省略号
+
+    # 创建横幅边框
     border_top = "╔" + "═" * (banner_width - 2) + "╗"
     border_bottom = "╚" + "═" * (banner_width - 2) + "╝"
 
     # 第一行标题
     title = "arXiv Pulse - 文献追踪系统"
-    # 标题居中
-    title_padding = (content_width - len(title) * 2) // 2  # 中文占2个英文字符宽度
-    if title_padding < 0:
-        title_padding = 0
-    title_line = "║" + " " * title_padding + title + " " * (content_width - len(title) * 2 - title_padding) + "║"
+    title_width = display_width(title)
+    # 计算左右填充
+    left_padding = (content_width - title_width) // 2
+    right_padding = content_width - title_width - left_padding
+    title_line = "║" + " " * left_padding + title + " " * right_padding + "║"
 
     # 第二行字段
-    # 简单处理：如果字段字符串太长，截断
-    max_field_len = content_width - 4  # 留出一些边距
-    if len(field_str) * 2 > max_field_len:  # 中文占2个英文字符宽度
-        # 截断字段字符串
-        field_str = field_str[: max_field_len // 2] + "..."
-
-    field_padding = (content_width - len(field_str) * 2) // 2
-    if field_padding < 0:
-        field_padding = 0
-    field_line = (
-        "║" + " " * field_padding + field_str + " " * (content_width - len(field_str) * 2 - field_padding) + "║"
-    )
+    # 最大字段显示宽度（留出边距）
+    max_field_width = content_width - 4
+    # 截断字段字符串如果太长
+    field_str = truncate_to_width(field_str, max_field_width)
+    field_width = display_width(field_str)
+
+    # 计算字段行的左右填充
+    left_padding = (content_width - field_width) // 2
+    right_padding = content_width - field_width - left_padding
+    field_line = "║" + " " * left_padding + field_str + " " * right_padding + "║"
 
     banner = f"\n{border_top}\n{title_line}\n{field_line}\n{border_bottom}\n"
     click.echo(banner)
 
 
-def sync_papers(years_back=1, summarize=False):
-    """同步论文（内部函数）"""
+def sync_papers(years_back=1, summarize=False, force=False):
+    """同步论文（内部函数）
+
+    Args:
+        years_back: 回溯的年数
+        summarize: 是否总结新论文
+        force: 是否强制同步（重新下载所有论文，忽略重复检查）
+    """
     crawler = ArXivCrawler()
     summarizer = PaperSummarizer()
 
-    click.echo(f"正在同步缺失论文（回溯 {years_back} 年）...")
+    mode_text = "强制同步" if force else "同步缺失论文"
+    click.echo(f"正在{mode_text}（回溯 {years_back} 年）...")
     click.echo("=" * 50)
 
     # 同步所有查询
     click.echo("1. 正在同步搜索查询...")
-    sync_result = crawler.sync_all_queries(years_back=years_back)
-    click.echo(f"   从查询添加了 {sync_result['total_new_papers']} 篇新论文")
+    sync_result = crawler.sync_all_queries(years_back=years_back, force=force)
+    result_text = "处理了" if force else "添加了"
+    click.echo(f"   从查询{result_text} {sync_result['total_new_papers']} 篇论文")
 
     # 同步重要论文
     click.echo("2. 正在同步重要论文...")
@@ -480,7 +502,7 @@ def sync_papers(years_back=1, summarize=False):
 
     click.echo("\n" + "=" * 50)
     click.echo("同步完成！")
-    click.echo(f"总共添加了新论文: {total_new}")
+    click.echo(f"总共{result_text}论文: {total_new}")
     click.echo(f"数据库现有 {crawl_stats['total_papers']} 篇论文")
     click.echo(f"已总结: {summary_stats['summarized_papers']} ({summary_stats['summarization_rate']:.1%})")
 
@@ -490,12 +512,13 @@ def sync_papers(years_back=1, summarize=False):
         "sync_result": sync_result,
         "important_result": important_result,
         "stats": {"crawl_stats": crawl_stats, "summary_stats": summary_stats},
+        "force_mode": force,
     }
 
 
 def get_workday_cutoff(days_back):
     """计算排除周末的截止日期"""
-    current = datetime.utcnow()
+    current = datetime.now(timezone.utc).replace(tzinfo=None)
     workdays_counted = 0
     days_to_go_back = 0
 
@@ -609,6 +632,11 @@ def generate_search_report(query, search_terms, papers, paper_limit=50, summariz
     """生成搜索结果的报告（内部函数）"""
     reporter = ReportGenerator()
 
+    # 如果没有找到论文，不生成报告
+    if not papers:
+        output.info("未找到论文，跳过报告生成")
+        return []
+
     # 设置报告限制
     original_limit = Config.REPORT_MAX_PAPERS
     Config.REPORT_MAX_PAPERS = paper_limit
@@ -698,10 +726,15 @@ def generate_search_report(query, search_terms, papers, paper_limit=50, summariz
 
 
 @click.group(context_settings={"help_option_names": ["-h", "--help"]})
+@click.option("--verbose", "-v", is_flag=True, help="显示详细输出（包括调试信息）")
 @click.version_option(version=__version__, prog_name="arXiv Pulse")
-def cli():
+def cli(verbose):
     """arXiv Pulse: 智能arXiv文献追踪和分析系统"""
-    pass
+    if verbose:
+        output.set_min_level(OutputLevel.DEBUG)
+    else:
+        # 确保使用环境变量中的LOG_LEVEL（output manager已经处理）
+        pass
 
 
 def interactive_configuration():
@@ -778,10 +811,10 @@ def interactive_configuration():
     click.echo("\n📊 爬虫配置")
     click.echo("-" * 40)
 
-    max_results_initial = click.prompt("初始同步每个查询的最大论文数", default=100, type=int, show_default=True)
+    max_results_initial = click.prompt("初始同步每个查询的最大论文数", default=10000, type=int, show_default=True)
     config["MAX_RESULTS_INITIAL"] = str(max_results_initial)
 
-    max_results_daily = click.prompt("每日同步每个查询的最大论文数", default=20, type=int, show_default=True)
+    max_results_daily = click.prompt("每日同步每个查询的最大论文数", default=500, type=int, show_default=True)
     config["MAX_RESULTS_DAILY"] = str(max_results_daily)
 
     years_back = click.prompt("初始同步回溯的年数", default=5, type=int, show_default=True)
@@ -866,27 +899,24 @@ def interactive_configuration():
     click.echo("-" * 40)
 
     # 根据领域数量提供建议
-    recommended_initial = 100
-    recommended_daily = 20
-
-    if num_selected_fields <= 3:
+    if num_selected_fields <= 6:
         click.echo("✅ 您选择了少量领域，保持默认配置即可。")
-    elif num_selected_fields <= 6:
-        recommended_initial = 70
-        recommended_daily = 15
+    elif num_selected_fields <= 10:
+        recommended_initial = 4000
+        recommended_daily = 200
         click.echo(f"⚠️  您选择了中等数量领域，建议调整爬虫配置以避免过多论文：")
-        click.echo(f"   - 初始同步每个查询最大论文数: {recommended_initial} (原默认: 100)")
-        click.echo(f"   - 每日同步每个查询最大论文数: {recommended_daily} (原默认: 20)")
+        click.echo(f"   - 初始同步每个查询最大论文数: {recommended_initial}")
+        click.echo(f"   - 每日同步每个查询最大论文数: {recommended_daily}")
     else:
-        recommended_initial = 50
-        recommended_daily = 10
+        recommended_initial = 1000
+        recommended_daily = 50
         click.echo(f"⚠️  您选择了大量领域 ({num_selected_fields}个)，强烈建议调整爬虫配置：")
-        click.echo(f"   - 初始同步每个查询最大论文数: {recommended_initial} (原默认: 100)")
-        click.echo(f"   - 每日同步每个查询最大论文数: {recommended_daily} (原默认: 20)")
+        click.echo(f"   - 初始同步每个查询最大论文数: {recommended_initial}")
+        click.echo(f"   - 每日同步每个查询最大论文数: {recommended_daily}")
         click.echo(f"   - 注意：同步大量领域可能需要较长时间和更多存储空间。")
 
     # 询问用户是否应用建议
-    if num_selected_fields > 3:
+    if num_selected_fields > 6:
         if click.confirm("\n💡 是否应用上述建议调整爬虫配置？", default=True):
             config["MAX_RESULTS_INITIAL"] = str(recommended_initial)
             config["MAX_RESULTS_DAILY"] = str(recommended_daily)
@@ -903,9 +933,6 @@ def interactive_configuration():
     report_max_papers = click.prompt("每份报告显示的最大论文数", default=50, type=int, show_default=True)
     config["REPORT_MAX_PAPERS"] = str(report_max_papers)
 
-    summary_sentences_limit = click.prompt("摘要句子数限制", default=3, type=int, show_default=True)
-    config["SUMMARY_SENTENCES_LIMIT"] = str(summary_sentences_limit)
-
     click.echo("\n✅ 配置完成！")
     return config, int(years_back)
 
@@ -920,7 +947,6 @@ def init(directory, years_back):
     # 创建目录结构
     (directory / "data").mkdir(exist_ok=True)
     (directory / "reports").mkdir(exist_ok=True)
-    (directory / "logs").mkdir(exist_ok=True)
 
     # 创建 .env 文件（如果不存在）
     env_file = directory / ".env"
@@ -937,59 +963,77 @@ def init(directory, years_back):
         if years_back is None:
             years_back = interactive_years_back
 
-        # 生成 .env 文件内容
-        env_content = f"""# arXiv Pulse 配置文件
-# 由交互式配置向导于 {datetime.now().strftime("%Y-%m-%d %H:%M:%S")} 生成
-
-# ========================
-# AI API 配置 (支持 OpenAI 格式)
-# ========================
-AI_API_KEY={config.get("AI_API_KEY", "your_api_key_here")}
-AI_MODEL={config.get("AI_MODEL", "DeepSeek-V3.2-Thinking")}
-AI_BASE_URL={config.get("AI_BASE_URL", "https://llmapi.paratera.com")}
-
-# ========================
-# 数据库配置
-# ========================
-DATABASE_URL=sqlite:///data/arxiv_papers.db
-
-# ========================
-# 爬虫配置
-# ========================
-MAX_RESULTS_INITIAL={config.get("MAX_RESULTS_INITIAL", "100")}    # init命令每个查询的论文数
-MAX_RESULTS_DAILY={config.get("MAX_RESULTS_DAILY", "20")}        # sync命令每个查询的论文数
-
-# ========================
-# 搜索查询配置
-# ========================
-# 分号分隔，允许查询中包含逗号
-# 根据您的选择生成的研究领域查询
-SEARCH_QUERIES={config.get("SEARCH_QUERIES", 'condensed matter physics AND cat:cond-mat.*; (ti:"density functional" OR abs:"density functional") AND (cat:physics.comp-ph OR cat:cond-mat.mtrl-sci OR cat:physics.chem-ph); (ti:"machine learning" OR abs:"machine learning") AND (cat:physics.comp-ph OR cat:cond-mat.mtrl-sci OR cat:physics.chem-ph)')}
-
-# ========================
-# 报告配置
-# ========================
-REPORT_DIR=reports
-SUMMARY_MAX_TOKENS=2000          # 总结和翻译的最大token数
-SUMMARY_SENTENCES_LIMIT={config.get("SUMMARY_SENTENCES_LIMIT", "3")}
-TOKEN_PRICE_PER_MILLION=3.0
-REPORT_MAX_PAPERS={config.get("REPORT_MAX_PAPERS", "50")}
-
-# ========================
-# 同步配置
-# ========================
-YEARS_BACK={config.get("YEARS_BACK", "3")}               # 同步回溯的年数
-IMPORTANT_PAPERS_FILE=important_papers.txt
-
-# ========================
-# 可选配置
-# ========================
-# 日志级别: DEBUG, INFO, WARNING, ERROR (默认: INFO)
-LOG_LEVEL=INFO
-
-# 爬虫延迟（秒，避免频繁请求 arXiv API）
-CRAWL_DELAY=1.0
-"""
+        # 读取 .ENV.TEMPLATE 文件作为基础模板
+        template_file = Path(__file__).parent / ".ENV.TEMPLATE"
+        if not template_file.exists():
+            click.echo(f"❌ 找不到模板文件: {template_file}")
+            click.echo("请确保 .ENV.TEMPLATE 文件存在于 arxiv_pulse 目录中")
+            return
+
+        env_content = template_file.read_text(encoding="utf-8")
+
+        # 添加生成时间戳注释（插入到第一行之后）
+        timestamp_comment = f"# 由交互式配置向导于 {datetime.now().strftime('%Y-%m-%d %H:%M:%S')} 生成\n"
+        lines = env_content.split("\n")
+        if lines and lines[0].startswith("#"):
+            # 在第一行注释后插入时间戳
+            lines.insert(1, timestamp_comment)
+        else:
+            # 如果没有注释行，添加到开头
+            lines.insert(0, timestamp_comment)
+        env_content = "\n".join(lines)
+
+        # 替换配置项（使用 config 字典中的值）
+        # 统一替换逻辑：搜索以键名开头的行，替换整行
+        lines = env_content.split("\n")
+
+        # AI API 配置
+        for i, line in enumerate(lines):
+            if line.strip().startswith("AI_API_KEY="):
+                lines[i] = f"AI_API_KEY={config.get('AI_API_KEY', 'your_api_key_here')}"
+                break
+
+        for i, line in enumerate(lines):
+            if line.strip().startswith("AI_MODEL="):
+                lines[i] = f"AI_MODEL={config.get('AI_MODEL', 'DeepSeek-V3.2-Thinking')}"
+                break
+
+        for i, line in enumerate(lines):
+            if line.strip().startswith("AI_BASE_URL="):
+                lines[i] = f"AI_BASE_URL={config.get('AI_BASE_URL', 'https://llmapi.paratera.com')}"
+                break
+
+        # 爬虫配置
+        for i, line in enumerate(lines):
+            if line.strip().startswith("MAX_RESULTS_INITIAL="):
+                lines[i] = f"MAX_RESULTS_INITIAL={config.get('MAX_RESULTS_INITIAL', '10000')}"
+                break
+
+        for i, line in enumerate(lines):
+            if line.strip().startswith("MAX_RESULTS_DAILY="):
+                lines[i] = f"MAX_RESULTS_DAILY={config.get('MAX_RESULTS_DAILY', '500')}"
+                break
+
+        # 搜索查询配置
+        default_search_queries = 'condensed matter physics AND cat:cond-mat.*; (ti:"density functional" OR abs:"density functional") AND (cat:physics.comp-ph OR cat:cond-mat.mtrl-sci OR cat:physics.chem-ph); (ti:"machine learning" OR abs:"machine learning") AND (cat:physics.comp-ph OR cat:cond-mat.mtrl-sci OR cat:physics.chem-ph)'
+        for i, line in enumerate(lines):
+            if line.strip().startswith("SEARCH_QUERIES="):
+                lines[i] = f"SEARCH_QUERIES={config.get('SEARCH_QUERIES', default_search_queries)}"
+                break
+
+        # 报告配置
+        for i, line in enumerate(lines):
+            if line.strip().startswith("REPORT_MAX_PAPERS="):
+                lines[i] = f"REPORT_MAX_PAPERS={config.get('REPORT_MAX_PAPERS', '50')}"
+                break
+
+        # 同步配置
+        for i, line in enumerate(lines):
+            if line.strip().startswith("YEARS_BACK="):
+                lines[i] = f"YEARS_BACK={config.get('YEARS_BACK', '5')}"
+                break
+
+        env_content = "\n".join(lines)
 
         env_file.write_text(env_content)
         click.echo(f"\n✅ 已在 {directory} 创建 .env 配置文件")
@@ -1000,10 +1044,12 @@ CRAWL_DELAY=1.0
             years_back = 5  # 默认值
 
     # 创建 important_papers.txt（如果不存在）
-    important_file = directory / "important_papers.txt"
+    important_file = directory / Config.IMPORTANT_PAPERS_FILE
+    # 确保父目录存在
+    important_file.parent.mkdir(parents=True, exist_ok=True)
     if not important_file.exists():
         important_file.write_text("# 在此添加重要论文的arXiv ID，每行一个\n")
-        click.echo(f"✅ 已在 {directory} 创建 important_papers.txt 文件")
+        click.echo(f"✅ 已创建重要论文文件: {important_file}")
 
     # 设置环境并验证配置
     if not setup_environment(directory):
@@ -1046,10 +1092,15 @@ CRAWL_DELAY=1.0
 
 @cli.command()
 @click.argument("directory", type=click.Path(exists=True, file_okay=False), default=".")
-@click.option("--years-back", type=int, default=1, help="同步回溯的年数（默认：1年）")
+@click.option("--years-back", type=int, default=None, help="同步回溯的年数（默认：强制模式5年，普通模式1年）")
 @click.option("--summarize/--no-summarize", default=False, help="是否总结新论文（默认：否）")
-def sync(directory, years_back, summarize):
-    """同步最新论文到数据库"""
+@click.option("--force", is_flag=True, default=False, help="强制同步：重新下载最近N年的所有论文，忽略重复检查")
+def sync(directory, years_back, summarize, force):
+    """同步最新论文到数据库
+
+    强制模式(--force): 重新下载最近N年的所有论文，忽略重复检查，默认回溯5年。
+    普通模式: 只下载缺失的新论文，默认回溯1年。
+    """
     directory = Path(directory).resolve()
     click.echo(f"正在同步 arXiv Pulse 于 {directory}")
 
@@ -1058,8 +1109,13 @@ def sync(directory, years_back, summarize):
 
     print_banner()
 
+    # 设置默认years_back值
+    if years_back is None:
+        years_back = 5 if force else 1
+        click.echo(f"使用默认回溯年数: {years_back} 年")
+
     # 同步论文
-    sync_result = sync_papers(years_back=years_back, summarize=summarize)
+    sync_result = sync_papers(years_back=years_back, summarize=summarize, force=force)
 
     click.echo("\n" + "=" * 50)
     click.echo("同步完成！数据库已更新。")
@@ -1073,8 +1129,19 @@ def sync(directory, years_back, summarize):
 @click.option("--use-ai/--no-ai", default=True, help="是否使用AI理解自然语言查询（默认：是）")
 @click.option("--summarize/--no-summarize", default=True, help="是否自动总结未总结的论文（默认：是）")
 @click.option("--max-summarize", type=int, default=0, help="最大总结论文数（默认：0表示无限制）")
-def search(query, directory, limit, years_back, use_ai, summarize, max_summarize):
-    """智能搜索论文（支持自然语言查询）"""
+@click.option("--categories", "-c", multiple=True, help="包含的分类（可多次使用）")
+@click.option("--days-back", type=int, help="回溯天数（例如：30表示最近30天）")
+@click.option("--authors", "-a", multiple=True, help="作者姓名（可多次使用）")
+@click.option(
+    "--sort-by",
+    type=click.Choice(["published", "relevance_score", "title", "updated"]),
+    default="published",
+    help="排序字段",
+)
+def search(
+    query, directory, limit, years_back, use_ai, summarize, max_summarize, categories, days_back, authors, sort_by
+):
+    """智能搜索论文（支持自然语言查询和基本过滤）"""
     directory = Path(directory).resolve()
 
     if not setup_environment(directory):
@@ -1086,7 +1153,7 @@ def search(query, directory, limit, years_back, use_ai, summarize, max_summarize
     crawler = ArXivCrawler()
     if years_back > 0:
         click.echo(f"搜索前先同步最近 {years_back} 年论文...")
-        sync_result = sync_papers(years_back=years_back, summarize=False)
+        sync_result = sync_papers(years_back=years_back, summarize=False, force=False)
         crawler = sync_result["crawler"]
 
     click.echo(f"\n正在搜索: '{query}'")
@@ -1105,10 +1172,22 @@ def search(query, directory, limit, years_back, use_ai, summarize, max_summarize
             用户正在搜索arXiv物理/计算材料科学论文，查询是: "{query}"
             
             请将自然语言查询转换为适合arXiv搜索的关键词或短语。
-            考虑以下领域：凝聚态物理、密度泛函理论(DFT)、机器学习、力场、分子动力学、量子化学。
             
-            返回格式：JSON数组，包含最多5个搜索关键词/短语。
-            示例：["machine learning materials science", "density functional theory", "condensed matter physics"]
+            重要规则：
+            1. 如果查询已经是明确的搜索词（如"DeepH"、"deep learning Hamiltonian"、"DFT计算"），直接使用它，不要添加同义词
+            2. 如果查询包含专业术语、缩写或专有名词，保持原样作为主要搜索词
+            3. 仅当查询非常模糊或一般性时（如"机器学习在材料科学中的应用"），才生成1-2个相关关键词
+            4. 优先保持查询的原始意图，不要添加不相关的关键词
+            5. 对于英文查询，保持原样；对于中文查询，翻译为英文关键词
+            考虑以下领域：凝聚态物理、密度泛函理论(DFT)、机器学习、力场、分子动力学、量子化学、计算材料科学。
+            
+            返回格式：JSON数组，包含1-2个搜索关键词/短语。
+            示例：
+            - 查询"DeepH": ["DeepH"]
+            - 查询"deep learning Hamiltonian": ["deep learning Hamiltonian"]
+            - 查询"DFT计算": ["DFT"]
+            - 查询"分子动力学模拟": ["molecular dynamics simulation"]
+            - 查询"机器学习在材料科学中的应用": ["machine learning materials science"]
             
             只返回JSON数组，不要其他文本。
             """
@@ -1116,7 +1195,10 @@ def search(query, directory, limit, years_back, use_ai, summarize, max_summarize
             response = client.chat.completions.create(
                 model=Config.AI_MODEL,
                 messages=[
-                    {"role": "system", "content": "你是arXiv论文搜索助手，擅长将自然语言查询转换为学术搜索关键词。"},
+                    {
+                        "role": "system",
+                        "content": "你是arXiv论文搜索助手，擅长识别专业术语并将自然语言查询转换为学术搜索关键词。",
+                    },
                     {"role": "user", "content": ai_prompt},
                 ],
                 max_tokens=200,
@@ -1141,30 +1223,30 @@ def search(query, directory, limit, years_back, use_ai, summarize, max_summarize
     with crawler.db.get_session() as session:
         from arxiv_pulse.models import Paper
 
-        all_results = []
-        for term in search_terms:
-            papers = (
-                session.query(Paper)
-                .filter(
-                    Paper.title.contains(term)
-                    | Paper.abstract.contains(term)
-                    | Paper.categories.contains(term)
-                    | Paper.search_query.contains(term)
-                )
-                .order_by(Paper.published.desc())
-                .limit(limit)
-                .all()
-            )
-            all_results.extend(papers)
+        # 使用增强搜索引擎进行模糊搜索
+        search_engine = SearchEngine(session)
 
-        # 去重并排序
-        unique_papers = {}
-        for paper in all_results:
-            if paper.arxiv_id not in unique_papers:
-                unique_papers[paper.arxiv_id] = paper
+        # 将搜索词合并为一个查询（搜索引擎会处理单词拆分和同义词扩展）
+        combined_query = " ".join(search_terms)
+
+        filter_config = SearchFilter(
+            query=combined_query,
+            search_fields=["title", "abstract"],
+            categories=list(categories) if categories else None,
+            authors=list(authors) if authors else None,
+            author_match="contains",  # 默认使用包含匹配
+            days_back=days_back,
+            limit=limit * min(len(search_terms), 2),  # 扩大限制但最多2倍，避免过多结果
+            sort_by=sort_by,
+            sort_order="desc",
+            match_all=True,  # AND逻辑：匹配所有搜索词
+        )
+
+        # 执行搜索
+        papers_to_show = search_engine.search_papers(filter_config)
 
-        sorted_papers = sorted(unique_papers.values(), key=lambda p: p.published or datetime.min, reverse=True)
-        papers_to_show = sorted_papers[:limit]
+        # 确保不超过限制
+        papers_to_show = papers_to_show[:limit]
 
         click.echo(f"找到 {len(papers_to_show)} 篇论文:")
 
@@ -1219,7 +1301,7 @@ def recent(directory, limit, days_back, years_back, summarize, max_summarize):
     # 先同步论文
     if years_back > 0:
         click.echo(f"报告前先同步最近 {years_back} 年论文...")
-        sync_papers(years_back=years_back, summarize=False)
+        sync_papers(years_back=years_back, summarize=False, force=False)
 
     # 生成报告
     click.echo("\n" + "=" * 50)
@@ -1294,8 +1376,6 @@ def stat(directory):
     # 时间分布
     click.echo(f"\n📅 时间分布:")
     with crawler.db.get_session() as session:
-        from datetime import datetime, timedelta
-
         # 按年统计
         year_stats = {}
         for paper in papers:
@@ -1319,290 +1399,5 @@ def stat(directory):
     click.echo("统计完成 ✅")
 
 
-@cli.command()
-@click.argument("paper_id")
-@click.argument("directory", type=click.Path(exists=True, file_okay=False), default=".")
-@click.option("--limit", default=10, help="返回结果的最大数量（默认：10）")
-@click.option("--threshold", type=float, default=0.5, help="相似度阈值（0.0-1.0，默认：0.5）")
-@click.option("--years-back", type=int, default=0, help="搜索前同步回溯的年数（默认：0，不更新）")
-def similar(paper_id, directory, limit, threshold, years_back):
-    """查找与指定论文相似的论文"""
-    directory = Path(directory).resolve()
-
-    if not setup_environment(directory):
-        sys.exit(1)
-
-    print_banner()
-
-    # 如果需要，先同步最新论文
-    crawler = ArXivCrawler()
-    if years_back > 0:
-        click.echo(f"搜索前先同步最近 {years_back} 年论文...")
-        sync_result = sync_papers(years_back=years_back, summarize=False)
-        crawler = sync_result["crawler"]
-
-    click.echo(f"\n查找与论文 '{paper_id}' 相似的论文")
-    click.echo("=" * 50)
-
-    with crawler.db.get_session() as session:
-        # 创建搜索引擎
-        search_engine = SearchEngine(session)
-
-        # 查找相似论文
-        click.echo(f"正在查找相似度≥{threshold}的论文...")
-        similar_papers_with_scores = search_engine.search_similar_papers(paper_id, limit=limit, threshold=threshold)
-
-        if not similar_papers_with_scores:
-            click.echo("未找到相似论文。")
-            return
-
-        click.echo(f"找到 {len(similar_papers_with_scores)} 篇相似论文:")
-
-        # 提取paper列表用于报告生成
-        similar_papers = [paper for paper, _ in similar_papers_with_scores]
-
-        # 显示结果
-        for i, (paper, similarity) in enumerate(similar_papers_with_scores, 1):
-            authors = json.loads(paper.authors) if paper.authors else []
-            author_names = [a.get("name", "") for a in authors[:2]]
-            if len(authors) > 2:
-                author_names.append("等")
-
-            click.echo(f"\n{i}. {paper.title}")
-            click.echo(f"   相似度: {similarity:.2f}")
-            click.echo(f"   作者: {', '.join(author_names)}")
-            click.echo(f"   arXiv ID: {paper.arxiv_id}")
-            click.echo(f"   分类: {paper.categories}")
-            click.echo(f"   发布日期: {paper.published.strftime('%Y-%m-%d') if paper.published else 'N/A'}")
-
-        # 生成报告
-        click.echo("\n正在生成相似论文报告...")
-        report_files = generate_search_report(
-            f"与 {paper_id} 相似的论文", [f"similar to {paper_id}"], similar_papers, paper_limit=limit
-        )
-
-        click.echo(f"报告生成完成：")
-        for f in report_files:
-            click.echo(f"  - {f}")
-
-
-@cli.command()
-@click.argument("query")
-@click.argument("directory", type=click.Path(exists=True, file_okay=False), default=".")
-@click.option("--limit", default=20, help="返回结果的最大数量（默认：20）")
-@click.option("--years-back", type=int, default=0, help="搜索前同步回溯的年数（默认：0，不更新）")
-@click.option("--use-ai/--no-ai", default=True, help="是否使用AI理解自然语言查询（默认：是）")
-@click.option("--categories", "-c", multiple=True, help="包含的分类（可多次使用）")
-@click.option("--exclude-categories", "-ec", multiple=True, help="排除的分类（可多次使用）")
-@click.option("--primary-category", "-pc", help="主要分类")
-@click.option("--authors", "-a", multiple=True, help="作者姓名（可多次使用）")
-@click.option(
-    "--author-match",
-    type=click.Choice(["contains", "exact", "any"]),
-    default="contains",
-    help="作者匹配方式：contains（包含）、exact（精确）、any（任一）",
-)
-@click.option("--date-from", type=click.DateTime(formats=["%Y-%m-%d"]), help="起始日期（格式：YYYY-MM-DD）")
-@click.option("--date-to", type=click.DateTime(formats=["%Y-%m-%d"]), help="结束日期（格式：YYYY-MM-DD）")
-@click.option("--days-back", type=int, help="回溯天数（例如：30表示最近30天）")
-@click.option("--summarized-only/--no-summarized-only", default=False, help="仅显示已总结的论文")
-@click.option("--downloaded-only/--no-downloaded-only", default=False, help="仅显示已下载的论文")
-@click.option(
-    "--sort-by",
-    type=click.Choice(["published", "relevance_score", "title", "updated", "created_at"]),
-    default="published",
-    help="排序字段",
-)
-@click.option("--sort-order", type=click.Choice(["asc", "desc"]), default="desc", help="排序顺序")
-@click.option("--match-all/--match-any", default=False, help="匹配所有条件（AND逻辑）或任一条件（OR逻辑）")
-def search_advanced(
-    query,
-    directory,
-    limit,
-    years_back,
-    use_ai,
-    categories,
-    exclude_categories,
-    primary_category,
-    authors,
-    author_match,
-    date_from,
-    date_to,
-    days_back,
-    summarized_only,
-    downloaded_only,
-    sort_by,
-    sort_order,
-    match_all,
-):
-    """高级搜索论文（支持多字段过滤）"""
-    directory = Path(directory).resolve()
-
-    if not setup_environment(directory):
-        sys.exit(1)
-
-    print_banner()
-
-    # 如果需要，先同步最新论文
-    crawler = ArXivCrawler()
-    if years_back > 0:
-        click.echo(f"搜索前先同步最近 {years_back} 年论文...")
-        sync_result = sync_papers(years_back=years_back, summarize=False)
-        crawler = sync_result["crawler"]
-
-    click.echo(f"\n高级搜索: '{query}'")
-    click.echo("=" * 50)
-
-    search_terms = [query]
-
-    # 如果启用AI且配置了AI API密钥，尝试解析自然语言查询
-    if use_ai and Config.AI_API_KEY:
-        try:
-            import openai
-
-            client = openai.OpenAI(api_key=Config.AI_API_KEY, base_url=Config.AI_BASE_URL)
-
-            ai_prompt = f"""
-            用户正在搜索arXiv物理/计算材料科学论文，查询是: "{query}"
-            
-            请将自然语言查询转换为适合arXiv搜索的关键词或短语。
-            考虑以下领域：凝聚态物理、密度泛函理论(DFT)、机器学习、力场、分子动力学、量子化学。
-            
-            返回格式：JSON数组，包含最多5个搜索关键词/短语。
-            示例：["machine learning materials science", "density functional theory", "condensed matter physics"]
-            
-            只返回JSON数组，不要其他文本。
-            """
-
-            response = client.chat.completions.create(
-                model=Config.AI_MODEL,
-                messages=[
-                    {"role": "system", "content": "你是arXiv论文搜索助手，擅长将自然语言查询转换为学术搜索关键词。"},
-                    {"role": "user", "content": ai_prompt},
-                ],
-                max_tokens=200,
-                temperature=0.3,
-            )
-
-            ai_response = response.choices[0].message.content
-            try:
-                search_terms = json.loads(ai_response)
-                if isinstance(search_terms, list) and len(search_terms) > 0:
-                    click.echo(f"AI解析的搜索词: {', '.join(search_terms[:3])}")
-                    if len(search_terms) > 3:
-                        click.echo(f"  以及 {len(search_terms) - 3} 个其他关键词")
-            except:
-                # 如果AI响应不是有效JSON，使用原始查询
-                pass
-
-        except Exception as e:
-            click.echo(f"AI解析失败，使用原始查询: {e}")
-
-    # 使用增强搜索引擎
-    with crawler.db.get_session() as session:
-        # 创建搜索过滤器
-        filter_config = SearchFilter(
-            query=query,
-            categories=list(categories) if categories else None,
-            exclude_categories=list(exclude_categories) if exclude_categories else None,
-            primary_category=primary_category,
-            authors=list(authors) if authors else None,
-            author_match=author_match,
-            date_from=date_from,
-            date_to=date_to,
-            days_back=days_back,
-            summarized_only=summarized_only,
-            downloaded_only=downloaded_only,
-            limit=limit,
-            sort_by=sort_by,
-            sort_order=sort_order,
-            match_all=match_all,
-        )
-
-        # 创建搜索引擎
-        search_engine = SearchEngine(session)
-
-        # 执行搜索
-        click.echo(f"正在搜索...")
-        papers = search_engine.search_papers(filter_config)
-
-        if not papers:
-            click.echo("未找到匹配的论文。")
-            return
-
-        click.echo(f"找到 {len(papers)} 篇论文:")
-
-        # 显示简要结果
-        for i, paper in enumerate(papers[:5], 1):  # 只显示前5篇作为预览
-            authors_list = json.loads(paper.authors) if paper.authors else []
-            author_names = [a.get("name", "") for a in authors_list[:2]]
-            if len(authors_list) > 2:
-                author_names.append("等")
-
-            click.echo(f"\n{i}. {paper.title}")
-            click.echo(f"   作者: {', '.join(author_names)}")
-            click.echo(f"   arXiv ID: {paper.arxiv_id}")
-            click.echo(f"   分类: {paper.categories}")
-            click.echo(f"   发布日期: {paper.published.strftime('%Y-%m-%d') if paper.published else 'N/A'}")
-            click.echo(f"   总结状态: {'已总结' if paper.summarized else '未总结'}")
-
-        if len(papers) > 5:
-            click.echo(f"\n... 以及 {len(papers) - 5} 篇更多论文")
-
-        # 生成搜索报告
-        click.echo("\n正在生成搜索报告...")
-        files = generate_search_report(directory, query, search_terms, papers, paper_limit=limit)
-
-        click.echo(f"报告生成完成：")
-        for f in files:
-            click.echo(f"  - {f}")
-        click.echo(f"\n详细论文信息、中文翻译和PDF链接请查看生成的Markdown报告。")
-
-
-@cli.command()
-@click.argument("directory", type=click.Path(exists=True, file_okay=False), default=".")
-@click.option("--limit", default=10, help="显示的搜索查询数量（默认：10）")
-def search_history(directory, limit):
-    """显示搜索历史（按使用频率排序）"""
-    directory = Path(directory).resolve()
-
-    if not setup_environment(directory):
-        sys.exit(1)
-
-    print_banner()
-
-    crawler = ArXivCrawler()
-
-    click.echo("\n" + "=" * 50)
-    click.echo("搜索历史")
-    click.echo("=" * 50)
-
-    with crawler.db.get_session() as session:
-        # 创建搜索引擎
-        search_engine = SearchEngine(session)
-
-        # 获取搜索历史
-        click.echo(f"正在获取搜索历史...")
-        history = search_engine.get_search_history(limit=limit)
-
-        if not history:
-            click.echo("暂无搜索历史。")
-            return
-
-        click.echo(f"\n找到 {len(history)} 个搜索查询:")
-        click.echo("-" * 50)
-
-        for i, item in enumerate(history, 1):
-            last_used = item["last_used"].strftime("%Y-%m-%d") if item["last_used"] else "N/A"
-            click.echo(f"\n{i}. 查询: {item['query']}")
-            click.echo(f"   使用次数: {item['count']}")
-            click.echo(f"   最后使用: {last_used}")
-            if item["last_paper_id"]:
-                click.echo(f"   最后论文ID: {item['last_paper_id']}")
-
-        click.echo(f"\n💡 提示: 使用 'pulse search \"查询内容\" .' 重用搜索")
-        click.echo(f"     或 'pulse search-advanced \"查询内容\" . --categories 分类' 进行高级搜索")
-
-
 if __name__ == "__main__":
     cli()