arxiv-pulse 0.5.0__py3-none-any.whl

arxiv_pulse/report_generator.py

@@ -0,0 +1,768 @@
+import json
+import pandas as pd
+from datetime import datetime, timedelta
+import markdown
+from typing import Dict, List, Any, Optional
+import logging
+import os
+
+from arxiv_pulse.models import Database, Paper
+from arxiv_pulse.config import Config
+from arxiv_pulse.output_manager import output
+
+# 使用根日志记录器的配置（保留用于向后兼容）
+logger = logging.getLogger(__name__)
+
+
+class ReportGenerator:
+    def __init__(self):
+        self.db = Database()
+        self.config = Config
+        self.total_tokens_used = 0  # 总token使用量
+        self.total_cost = 0.0  # 总费用（元）
+        self.token_price_per_million = Config.TOKEN_PRICE_PER_MILLION  # 每百万token价格，可从配置覆盖
+        self.summary_sentences_limit = Config.SUMMARY_SENTENCES_LIMIT  # 摘要句子数限制
+
+        # 抑制第三方库的详细日志
+        import logging
+
+        logging.getLogger("httpx").setLevel(logging.WARNING)
+        logging.getLogger("httpcore").setLevel(logging.WARNING)
+
+        # arXiv分类解释映射
+        self.category_explanations = {
+            # Computer Science
+            "cs.AI": "人工智能 (Artificial Intelligence)",
+            "cs.CL": "计算语言学 (Computation and Language)",
+            "cs.CR": "密码学与安全 (Cryptography and Security)",
+            "cs.CV": "计算机视觉 (Computer Vision)",
+            "cs.LG": "机器学习 (Machine Learning)",
+            "cs.NE": "神经网络 (Neural and Evolutionary Computing)",
+            "cs.SE": "软件工程 (Software Engineering)",
+            "cs.PL": "编程语言 (Programming Languages)",
+            "cs.DC": "分布式计算 (Distributed, Parallel, and Cluster Computing)",
+            "cs.DS": "数据结构与算法 (Data Structures and Algorithms)",
+            "cs.IT": "信息论 (Information Theory)",
+            "cs.SY": "系统与控制 (Systems and Control)",
+            # Physics
+            "cond-mat": "凝聚态物理 (Condensed Matter)",
+            "cond-mat.mtrl-sci": "材料科学 (Materials Science)",
+            "cond-mat.str-el": "强关联电子系统 (Strongly Correlated Electrons)",
+            "cond-mat.supr-con": "超导 (Superconductivity)",
+            "cond-mat.mes-hall": "介观系统与量子霍尔效应 (Mesoscopic Systems and Quantum Hall Effect)",
+            "cond-mat.soft": "软凝聚态物质 (Soft Condensed Matter)",
+            "cond-mat.dis-nn": "无序系统与神经网络 (Disordered Systems and Neural Networks)",
+            "cond-mat.stat-mech": "统计力学 (Statistical Mechanics)",
+            "cond-mat.quant-gas": "量子气体 (Quantum Gases)",
+            # Physics other
+            "physics": "物理学 (Physics)",
+            "physics.comp-ph": "计算物理 (Computational Physics)",
+            "physics.chem-ph": "化学物理 (Chemical Physics)",
+            "physics.data-an": "数据分析 (Data Analysis, Statistics and Probability)",
+            "physics.ins-det": "仪器与探测器 (Instrumentation and Detectors)",
+            # Mathematics
+            "math": "数学 (Mathematics)",
+            "math.NA": "数值分析 (Numerical Analysis)",
+            "math.OC": "优化与控制 (Optimization and Control)",
+            "math.ST": "统计 (Statistics)",
+            # Quantitative Biology
+            "q-bio": "定量生物学 (Quantitative Biology)",
+            "q-bio.BM": "生物分子 (Biomolecules)",
+            "q-bio.QM": "定量方法 (Quantitative Methods)",
+            # Quantitative Finance
+            "q-fin": "定量金融 (Quantitative Finance)",
+            # Statistics
+            "stat": "统计学 (Statistics)",
+            "stat.ML": "机器学习 (Machine Learning)",
+            "stat.AP": "应用 (Applications)",
+            "stat.CO": "计算 (Computation)",
+            "stat.ME": "方法学 (Methodology)",
+            "stat.OT": "其他 (Other)",
+            "stat.TH": "理论 (Theory)",
+        }
+
+    def get_category_explanation(self, category_code: str) -> str:
+        """获取分类代码的解释"""
+        # 处理可能的多个分类
+        categories = [c.strip() for c in category_code.split(",")]
+        explanations = []
+
+        for cat in categories:
+            # 处理主分类，如cond-mat.mtrl-sci
+            if cat in self.category_explanations:
+                explanations.append(self.category_explanations[cat])
+            else:
+                # 尝试匹配主分类前缀
+                main_cat = cat.split(".")[0] if "." in cat else cat
+                if main_cat in self.category_explanations:
+                    explanations.append(f"{cat} ({self.category_explanations[main_cat].split('(')[0]})")
+                else:
+                    explanations.append(cat)
+
+        return "; ".join(explanations)
+
+    def clean_json_response(self, text):
+        """清理AI响应中的JSON代码块标记"""
+        import re
+
+        if not text:
+            return text
+        # 移除 ```json 和 ``` 标记
+        text = text.strip()
+        # 匹配 ```json ... ``` 模式
+        json_match = re.search(r"```json\s*(.*?)\s*```", text, re.DOTALL)
+        if json_match:
+            return json_match.group(1).strip()
+        # 匹配 ``` ... ``` 模式（没有json标签）
+        code_match = re.search(r"```\s*(.*?)\s*```", text, re.DOTALL)
+        if code_match:
+            return code_match.group(1).strip()
+        # 如果以 ```json 开头但没有闭合
+        if text.startswith("```json"):
+            text = text[7:].strip()
+        if text.startswith("```"):
+            text = text[3:].strip()
+        if text.endswith("```"):
+            text = text[:-3].strip()
+        return text
+
+    def calculate_relevance_score(self, paper) -> int:
+        """计算论文相关度评级 (1-5星)"""
+        # 基于搜索查询和分类的评分
+        query = paper.search_query or ""
+        categories = paper.categories or ""
+
+        # 核心研究领域 (凝聚态物理、DFT、力场等)
+        core_domains = [
+            "condensed matter physics",
+            "density functional theory",
+            "first principles calculation",
+            "force fields",
+            "molecular dynamics",
+            "computational materials science",
+            "quantum chemistry",
+        ]
+
+        # 相关领域 (机器学习等)
+        related_domains = ["machine learning"]
+
+        # 目标分类 (凝聚态物理、计算物理等)
+        target_categories = [
+            "cond-mat",  # 凝聚态物理
+            "physics.comp-ph",  # 计算物理
+            "physics.chem-ph",  # 化学物理
+            "quant-ph",  # 量子物理
+        ]
+
+        # 相关分类 (机器学习、人工智能)
+        related_categories = [
+            "cs.ai",
+            "cs.lg",
+            "stat.ml",
+            "cs.ne",  # 机器学习
+            "physics.data-an",  # 数据分析
+        ]
+
+        # 不相关分类 (网络安全、其他CS领域)
+        unrelated_categories = [
+            "cs.cr",  # 密码学与安全
+            "cs.pl",  # 编程语言
+            "cs.se",  # 软件工程
+            "cs.cv",  # 计算机视觉 (除非明确相关)
+            "cs.cl",  # 计算语言学
+        ]
+
+        query_lower = query.lower()
+        categories_lower = categories.lower()
+
+        # 评分逻辑
+        score = 3  # 默认3星
+
+        # 1. 检查搜索查询
+        for domain in core_domains:
+            if domain in query_lower:
+                score += 2  # 核心领域加分
+                break
+
+        for domain in related_domains:
+            if domain in query_lower:
+                score += 1  # 相关领域加分
+                break
+
+        # 2. 检查分类匹配
+        # 如果分类与目标领域匹配
+        if any(cat in categories_lower for cat in target_categories):
+            score += 2
+
+        # 如果分类与相关领域匹配
+        elif any(cat in categories_lower for cat in related_categories):
+            score += 1
+
+        # 如果分类明显不相关
+        if any(cat in categories_lower for cat in unrelated_categories):
+            score -= 1
+
+        # 3. 特殊情况：搜索查询是计算材料科学但分类是CS领域
+        if "computational materials science" in query_lower and any(
+            cat in categories_lower for cat in unrelated_categories
+        ):
+            score = max(2, score - 1)  # 降低评分
+
+        # 确保分数在1-5之间
+        return max(1, min(5, score))
+
+    def _truncate_to_sentences(self, text: str, max_sentences: Optional[int] = None) -> str:
+        """将文本截断为指定数量的句子（支持中英文）"""
+        if not text:
+            return ""
+
+        if max_sentences is None:
+            max_sentences = self.summary_sentences_limit
+
+        import re
+
+        # 支持中英文句子分隔符：句号、问号、感叹号、分号、省略号
+        # 英文: . ? ! ; ... 中文: 。！？；…
+        pattern = r"([。！？；…\.\?!;]+|\.{3,})"
+        parts = re.split(pattern, text)
+
+        sentences = []
+        current = ""
+        for i, part in enumerate(parts):
+            current += part
+            if i % 2 == 1:  # 分隔符部分
+                sentences.append(current)
+                current = ""
+
+        # 如果最后还有未结束的句子
+        if current:
+            sentences.append(current)
+
+        # 如果分割失败，按长度简单截断
+        if len(sentences) == 0:
+            return text[:200] + "..." if len(text) > 200 else text
+
+        # 取前max_sentences句
+        result = "".join(sentences[:max_sentences])
+
+        # 如果截断后比原文本短很多，添加省略号
+        if len(result) < len(text) * 0.8:
+            # 移除末尾的句子分隔符，添加省略号
+            result = result.rstrip("。！？；….?!;") + "…"
+
+        return result
+
+    def translate_text(self, text: str, target_lang: str = "zh") -> str:
+        """使用DeepSeek或OpenAI API翻译文本，优先使用缓存"""
+        if not text or not text.strip():
+            return ""
+
+        try:
+            import openai
+
+            # 1. 检查缓存
+            cached_translation = self.db.get_translation_cache(text, target_lang)
+            if cached_translation:
+                output.info(f"✓ 缓存命中 ({len(text)} 字符)")
+                return cached_translation
+
+            # 2. 使用DeepSeek
+            if Config.AI_API_KEY:
+                translated = self._translate_with_deepseek(text, target_lang)
+                # 3. 存储到缓存
+                if translated and not translated.startswith("*"):
+                    self.db.set_translation_cache(text, translated, target_lang)
+                return translated
+            else:
+                return "*翻译需要配置DeepSeek API密钥*"
+
+        except ImportError:
+            return "*需要安装openai库*"
+        except Exception as e:
+            output.error("翻译文本失败", details={"exception": str(e)})
+            return f"*翻译出错: {str(e)[:100]}*"
+
+    def _translate_with_deepseek(self, text: str, target_lang: str = "zh") -> str:
+        """使用DeepSeek API翻译文本"""
+        import openai
+
+        # 配置DeepSeek (openai 2.x版本)
+        client = openai.OpenAI(api_key=Config.AI_API_KEY, base_url=Config.AI_BASE_URL)
+
+        # 如果文本太长，截断（API有token限制）
+        max_chars = 3000
+        if len(text) > max_chars:
+            text_to_translate = text[:max_chars] + "... [文本过长，已截断]"
+        else:
+            text_to_translate = text
+
+        # 准备翻译提示
+        if target_lang == "zh":
+            system_prompt = "你是一个专业的翻译助手。将以下英文文本翻译成中文，保持专业术语准确，语言流畅。"
+        else:
+            system_prompt = f"你是一个专业的翻译助手。将以下文本翻译成{target_lang}，保持专业术语准确，语言流畅。"
+
+        try:
+            response = client.chat.completions.create(
+                model=Config.AI_MODEL or "DeepSeek-V3.2-Thinking",
+                messages=[
+                    {"role": "system", "content": system_prompt},
+                    {"role": "user", "content": text_to_translate},
+                ],
+                max_tokens=min(2000, len(text_to_translate) // 2),
+                temperature=0.3,
+            )
+
+            # 记录token使用情况并更新统计
+            if hasattr(response, "usage") and response.usage:
+                usage = response.usage
+                current_tokens = usage.total_tokens
+                self.total_tokens_used += current_tokens
+
+                # DeepSeek实际价格：输入0.14元/百万token，输出0.28元/百万token
+                # 如果用户指定了价格则使用用户价格，否则使用实际价格
+                if self.token_price_per_million is not None:
+                    # 使用用户指定的统一价格
+                    cost = (current_tokens / 1_000_000) * self.token_price_per_million
+                else:
+                    # 使用DeepSeek实际价格
+                    input_cost = (usage.prompt_tokens / 1_000_000) * 0.14
+                    output_cost = (usage.completion_tokens / 1_000_000) * 0.28
+                    cost = input_cost + output_cost
+
+                # 更新累计总费用
+                self.total_cost += cost
+                total_cost_formatted = f"{self.total_cost:.4f}"
+                output.info(
+                    f"✓ 翻译完成 | 本次: {current_tokens} tokens ({cost:.4f}¥) | 累计: {self.total_tokens_used} tokens (¥{total_cost_formatted})"
+                )
+            else:
+                # 估算token使用（约4字符/1token）
+                estimated_tokens = len(text) // 4 + 500  # 基础估计
+                self.total_tokens_used += estimated_tokens
+                # 计算当前批次费用
+                price_per_million = (
+                    self.token_price_per_million if self.token_price_per_million is not None else 0.21
+                )  # DeepSeek平均价格
+                current_cost = (estimated_tokens / 1_000_000) * price_per_million
+                self.total_cost += current_cost
+                total_cost_formatted = f"{self.total_cost:.4f}"
+                output.info(
+                    f"✓ 翻译完成 | 本次: ~{estimated_tokens} tokens ({current_cost:.4f}¥) | 累计: {self.total_tokens_used} tokens (¥{total_cost_formatted})"
+                )
+
+            translated = response.choices[0].message.content
+            return translated.strip() if translated else ""
+
+        except Exception as e:
+            output.error("DeepSeek翻译失败", details={"exception": str(e)})
+            raise e
+
+    def generate_daily_report(self) -> Dict[str, Any]:
+        """Generate daily report of new papers"""
+        output.do("生成每日报告")
+
+        with self.db.get_session() as session:
+            # Get papers from last 24 hours
+            cutoff = datetime.utcnow() - timedelta(hours=24)
+            new_papers = (
+                session.query(Paper)
+                .filter(Paper.created_at >= cutoff)
+                .order_by(Paper.published.desc())
+                .limit(self.config.REPORT_MAX_PAPERS)
+                .all()
+            )
+
+            # Get summarized papers
+            summarized = [p for p in new_papers if getattr(p, "summarized", False) == True]
+
+            # Group by category/query
+            by_query = {}
+            for paper in new_papers:
+                query = paper.search_query or "Unknown"
+                if query not in by_query:
+                    by_query[query] = []
+                by_query[query].append(paper)
+
+            # Statistics
+            stats = {
+                "total_new": len(new_papers),
+                "summarized_new": len(summarized),
+                "papers_by_query": {k: len(v) for k, v in by_query.items()},
+                "date_generated": datetime.now().isoformat(),
+                "report_type": "daily",
+            }
+
+            return {
+                "stats": stats,
+                "papers": new_papers,
+                "summarized_papers": summarized,
+                "grouped_papers": by_query,
+            }
+
+    def generate_weekly_report(self) -> Dict[str, Any]:
+        """Generate weekly report"""
+        output.do("生成每周报告")
+
+        with self.db.get_session() as session:
+            # Get papers from last 7 days
+            cutoff = datetime.utcnow() - timedelta(days=7)
+            recent_papers = (
+                session.query(Paper)
+                .filter(Paper.created_at >= cutoff)
+                .order_by(Paper.published.desc())
+                .limit(self.config.REPORT_MAX_PAPERS)
+                .all()
+            )
+
+            # Get database stats
+            db_stats = self.db.get_statistics()
+
+            # Top categories
+            categories = db_stats.get("categories_distribution", {})
+            top_categories = sorted(categories.items(), key=lambda x: x[1], reverse=True)[:10]
+
+            # Statistics
+            stats = {
+                "total_recent": len(recent_papers),
+                "database_stats": db_stats,
+                "top_categories": dict(top_categories),
+                "date_generated": datetime.now().isoformat(),
+                "report_type": "weekly",
+            }
+
+            return {
+                "stats": stats,
+                "recent_papers": recent_papers,
+                "database_stats": db_stats,
+            }
+
+    def save_markdown_report(self, report_data: Dict[str, Any], filename: Optional[str] = None) -> str:
+        """Save report as markdown file"""
+        if filename is None:
+            timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+            report_type = report_data.get("stats", {}).get("report_type", "report")
+            filename = f"{report_type}_{timestamp}.md"
+
+        filepath = os.path.join(self.config.REPORT_DIR, filename)
+
+        # 重置token统计（每次新报告开始时）
+        self.total_tokens_used = 0
+        self.total_cost = 0.0
+        output.info("开始生成报告 - token计数已重置")
+
+        # Generate markdown content
+        stats = report_data["stats"]
+
+        # 报告类型中文映射
+        report_type_chinese = {"daily": "每日", "weekly": "每周", "recent": "最近", "search": "搜索"}
+        report_type = report_type_chinese.get(stats["report_type"], stats["report_type"])
+
+        markdown_content = f"""# arXiv 文献报告
+**生成时间**: {datetime.now().strftime("%Y-%m-%d %H:%M:%S")}
+**报告类型**: {report_type}报告
+
+## 统计摘要
+"""
+
+        if stats["report_type"] == "daily":
+            markdown_content += f"""
+- **今日新论文**: {stats["total_new"]}
+- **今日已总结**: {stats["summarized_new"]}
+- **总结率**: {stats["summarized_new"] / stats["total_new"]:.1%} (如果总数 > 0)
+
+### 按搜索查询统计
+"""
+            for query, count in stats["papers_by_query"].items():
+                markdown_content += f"- **{query}**: {count} 篇论文\n"
+
+        elif stats["report_type"] == "weekly":
+            markdown_content += f"""
+- **本周论文**: {stats["total_recent"]}
+- **数据库总论文**: {stats["database_stats"]["total_papers"]}
+- **已总结论文**: {stats["database_stats"]["summarized_papers"]}
+- **总体总结率**: {stats["database_stats"]["summarized_papers"] / stats["database_stats"]["total_papers"]:.1%} (如果总数 > 0)
+
+### 热门分类
+"""
+            for category, count in stats["top_categories"].items():
+                markdown_content += f"- **{category}**: {count} 篇论文\n"
+
+        elif stats["report_type"] == "recent":
+            markdown_content += f"""
+- **最近论文**: {stats["total_recent"]} (最近 {stats["days_back"]} 天)
+- **数据库总论文**: {stats.get("database_stats", {}).get("total_papers", "N/A")}
+- **已总结论文**: {stats.get("database_stats", {}).get("summarized_papers", "N/A")}
+
+### 热门分类
+"""
+            for category, count in stats.get("top_categories", {}).items():
+                markdown_content += f"- **{category}**: {count} 篇论文\n"
+
+        elif stats["report_type"] == "search":
+            markdown_content += f"""
+- **搜索查询**: {stats.get("original_query", "N/A")}
+- **找到论文**: {stats.get("total_found", "N/A")}
+- **数据库总论文**: {stats.get("database_stats", {}).get("total_papers", "N/A")}
+- **已总结论文**: {stats.get("database_stats", {}).get("summarized_papers", "N/A")}
+
+### 搜索词
+"""
+            search_terms = stats.get("search_terms", [])
+            if isinstance(search_terms, list):
+                for term in search_terms[:5]:  # 只显示前5个搜索词
+                    markdown_content += f"- **{term}**\n"
+            else:
+                markdown_content += f"- **{search_terms}**\n"
+
+            markdown_content += "\n### 热门分类\n"
+            for category, count in stats.get("top_categories", {}).items():
+                markdown_content += f"- **{category}**: {count} 篇论文\n"
+
+        # Add paper details
+        if "papers" in report_data and report_data["papers"]:
+            markdown_content += "\n## 新论文\n\n"
+
+            papers = report_data["papers"][:50]  # Limit to 50 papers
+            total_papers = len(papers)
+
+            for i, paper in enumerate(papers, 1):
+                try:
+                    output.do(f"[{i}/{total_papers}] 处理论文: {paper.arxiv_id}")
+                    authors = json.loads(paper.authors) if paper.authors else []
+                    author_names = [a.get("name", "") for a in authors[:3]]
+                    if len(authors) > 3:
+                        author_names.append("et al.")
+
+                    # 计算相关度评级
+                    relevance_score = self.calculate_relevance_score(paper)
+                    stars = "★" * relevance_score + "☆" * (5 - relevance_score)
+
+                    # 题目中英双语
+                    # 尝试翻译标题
+                    output.do(f"[{i}/{total_papers}] 翻译标题")
+                    title_translation = self.translate_text(paper.title, "zh")
+
+                    # 根据用户要求：中文标题为主，英文标题放在括号里
+                    if title_translation and not title_translation.startswith("*"):
+                        markdown_content += f"### {title_translation} ({paper.title})\n\n"
+                    else:
+                        markdown_content += f"### {paper.title}\n\n"
+
+                    markdown_content += f"**作者 (Authors)**: {', '.join(author_names)}\n"
+                    markdown_content += f"**发表日期 (Published)**: {paper.published.strftime('%Y-%m-%d') if paper.published else 'N/A'}\n"
+
+                    # 分类解释
+                    category_explanation = self.get_category_explanation(paper.categories or "")
+                    markdown_content += f"**分类解释 (Categories)**: {category_explanation}\n"
+                    markdown_content += f"**原始分类代码 (Original Codes)**: {paper.categories}\n"
+
+                    markdown_content += f"**搜索查询 (Search Query)**: {paper.search_query}\n"
+
+                    # 相关度评级
+                    markdown_content += f"**相关度评级 (Relevance)**: {stars} ({relevance_score}/5)\n\n"
+
+                    # 显示关键发现（如果存在）
+                    if paper.summary:
+                        summary_data = None
+
+                        # 尝试解析总结JSON
+                        try:
+                            summary_data = json.loads(paper.summary)
+                        except json.JSONDecodeError:
+                            # 如果直接解析失败，尝试清理后解析
+                            cleaned_summary = self.clean_json_response(paper.summary)
+                            try:
+                                summary_data = json.loads(cleaned_summary)
+                            except json.JSONDecodeError:
+                                # 如果仍然不是JSON，忽略总结数据
+                                pass
+
+                        # 显示关键发现（如果存在）
+                        if summary_data and "key_findings" in summary_data and summary_data["key_findings"]:
+                            markdown_content += "**关键发现 (Key Findings)**:\n"
+                            for finding in summary_data["key_findings"][:5]:
+                                markdown_content += f"- {finding}\n"
+                            markdown_content += "\n"
+
+                    # 完整英文摘要和中文翻译
+                    if paper.abstract:
+                        markdown_content += f"**完整英文摘要 (Full Abstract)**:\n{paper.abstract}\n\n"
+
+                        # 尝试翻译摘要
+                        output.do(f"[{i}/{total_papers}] 翻译摘要")
+                        chinese_translation = self.translate_text(paper.abstract, "zh")
+                        if chinese_translation and not chinese_translation.startswith("*"):
+                            markdown_content += f"**中文翻译 (Chinese Translation)**:\n{chinese_translation}\n\n"
+                        elif chinese_translation:
+                            markdown_content += f"**中文翻译 (Chinese Translation)**: {chinese_translation}\n\n"
+                        else:
+                            markdown_content += f"**中文翻译 (Chinese Translation)**: *翻译服务不可用*\n\n"
+                    else:
+                        markdown_content += f"**摘要 (Abstract)**: 无摘要\n\n"
+
+                    markdown_content += f"**arXiv ID**: [{paper.arxiv_id}](https://arxiv.org/abs/{paper.arxiv_id})\n"
+                    markdown_content += f"**PDF**: [下载 (Download)]({paper.pdf_url})\n\n"
+                    markdown_content += "---\n\n"
+
+                except Exception as e:
+                    output.error(
+                        f"格式化论文失败: {paper.arxiv_id}",
+                        details={"exception": str(e)},
+                    )
+                    continue
+
+        # Add recommendations section
+        markdown_content += "\n## 建议\n\n"
+
+        if stats["report_type"] == "daily":
+            markdown_content += "1. 浏览您感兴趣领域的新论文\n"
+            markdown_content += "2. 查看已总结的论文以快速了解内容\n"
+            markdown_content += "3. 将相关论文添加到阅读列表\n"
+        else:
+            markdown_content += "1. 回顾您研究领域的每周趋势\n"
+            markdown_content += "2. 从热门分类中识别新兴主题\n"
+            markdown_content += "3. 规划下周的阅读计划\n"
+
+        # Save to file
+        with open(filepath, "w", encoding="utf-8") as f:
+            f.write(markdown_content)
+
+        # 显示最终token统计
+        if self.total_tokens_used > 0:
+            total_cost_formatted = f"{self.total_cost:.4f}"
+            output.info(f"报告生成完成 - 总计: {self.total_tokens_used} tokens (¥{total_cost_formatted})")
+
+        output.done(f"报告已保存: {filepath}")
+        return filepath
+
+    def save_csv_report(self, report_data: Dict[str, Any], filename: Optional[str] = None) -> Optional[str]:
+        """Save report as CSV file"""
+        if filename is None:
+            timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+            report_type = report_data["stats"]["report_type"]
+            filename = f"{report_type}_report_{timestamp}.csv"
+
+        filepath = os.path.join(self.config.REPORT_DIR, filename)
+
+        # Convert papers to DataFrame
+        papers_data = []
+        for paper in report_data.get("papers", []):
+            try:
+                papers_data.append(
+                    {
+                        "arxiv_id": paper.arxiv_id,
+                        "title": paper.title,
+                        "authors": paper.authors,
+                        "published": paper.published.isoformat() if paper.published else None,
+                        "categories": paper.categories,
+                        "primary_category": paper.primary_category,
+                        "search_query": paper.search_query,
+                        "summarized": paper.summarized,
+                        "pdf_url": paper.pdf_url,
+                        "doi": paper.doi,
+                        "created_at": paper.created_at.isoformat() if paper.created_at else None,
+                    }
+                )
+            except Exception as e:
+                output.error(
+                    f"处理论文到CSV失败: {paper.arxiv_id}",
+                    details={"exception": str(e)},
+                )
+
+        if papers_data:
+            df = pd.DataFrame(papers_data)
+
+            # 中文列名映射
+            chinese_columns = {
+                "arxiv_id": "arXiv ID",
+                "title": "标题",
+                "authors": "作者",
+                "published": "发表日期",
+                "categories": "分类",
+                "primary_category": "主要分类",
+                "search_query": "搜索查询",
+                "summarized": "已总结",
+                "pdf_url": "PDF链接",
+                "doi": "DOI",
+                "created_at": "创建时间",
+            }
+
+            # 重命名列
+            df = df.rename(columns=chinese_columns)
+
+            df.to_csv(filepath, index=False, encoding="utf-8")
+            output.done(f"CSV报告已保存: {filepath}")
+            return filepath
+        else:
+            output.warn("没有论文数据可保存为CSV")
+            return None
+
+    def generate_and_save_daily_report(self) -> List[str]:
+        """Generate and save daily report (returns list of saved files)"""
+        report_data = self.generate_daily_report()
+
+        saved_files = []
+
+        # Save markdown report
+        md_file = self.save_markdown_report(report_data)
+        if md_file:
+            saved_files.append(md_file)
+
+        # Save CSV report
+        csv_file = self.save_csv_report(report_data)
+        if csv_file:
+            saved_files.append(csv_file)
+
+        return saved_files
+
+    def generate_and_save_weekly_report(self) -> List[str]:
+        """Generate and save weekly report"""
+        report_data = self.generate_weekly_report()
+
+        saved_files = []
+
+        # Save markdown report
+        md_file = self.save_markdown_report(report_data)
+        if md_file:
+            saved_files.append(md_file)
+
+        # Save CSV report
+        csv_file = self.save_csv_report(report_data)
+        if csv_file:
+            saved_files.append(csv_file)
+
+        return saved_files
+
+
+def main():
+    """Test report generator"""
+    generator = ReportGenerator()
+
+    print("Testing report generator...")
+
+    # Generate daily report
+    print("\nGenerating daily report...")
+    daily_data = generator.generate_daily_report()
+    print(f"Daily stats: {daily_data['stats']}")
+
+    # Save reports
+    print("\nSaving reports...")
+    saved_files = generator.generate_and_save_daily_report()
+    print(f"Saved files: {saved_files}")
+
+    # Generate weekly report
+    print("\nGenerating weekly report...")
+    weekly_data = generator.generate_weekly_report()
+    print(f"Weekly stats: {weekly_data['stats'].get('total_recent', 0)} recent papers")
+
+    # Check report directory
+    report_dir = Config.REPORT_DIR
+    print(f"\nReport directory: {report_dir}")
+    if os.path.exists(report_dir):
+        files = os.listdir(report_dir)
+        print(f"Existing reports: {len(files)} files")
+
+
+if __name__ == "__main__":
+    main()