feat: 军事科技每日资讯推送系统 - Docker部署 + 日志系统 + 数据目录重组

scripts/modules/monitor.py

@@ -0,0 +1,162 @@
+# -*- coding: utf-8 -*-
+"""
+运行时监控模块
+统计各环节耗时、API调用量/token消耗、成功率，运行结束输出统计报表
+"""
+import time
+from collections import defaultdict
+from contextlib import contextmanager
+from .logger import get_logger
+
+logger = get_logger()
+
+STAGE_ORDER = [
+    "rss_fetch", "time_filter", "deduplicate", "translate_titles",
+    "keyword_filter", "translate_contents", "ai_analyze",
+    "webzine_generate", "category_overview", "report_generate", "image_generate"
+]
+
+STAGE_LABELS = {
+    "rss_fetch": "RSS抓取",
+    "time_filter": "时间过滤",
+    "deduplicate": "去重",
+    "translate_titles": "标题翻译",
+    "keyword_filter": "关键词筛选",
+    "translate_contents": "正文翻译",
+    "ai_analyze": "AI评分分类",
+    "webzine_generate": "网摘生成",
+    "category_overview": "分类介绍",
+    "report_generate": "报告生成",
+    "image_generate": "图片生成",
+}
+
+
+class RunMonitor:
+    def __init__(self):
+        self.start_time = time.time()
+        self.stage_times = {}
+        self.stage_active = {}
+        self.api_calls = []
+        self.source_success = defaultdict(int)
+        self.source_failure = defaultdict(int)
+        self.articles_total = 0
+        self.articles_cached = 0
+        self.articles_processed = 0
+
+    @contextmanager
+    def stage(self, name):
+        """上下文管理器，自动记录阶段耗时"""
+        start = time.perf_counter()
+        try:
+            yield
+        finally:
+            elapsed = time.perf_counter() - start
+            self.stage_times[name] = elapsed
+
+    def record_api_call(self, model, purpose, max_tokens, success, retries=0):
+        """记录一次API调用"""
+        self.api_calls.append({
+            "model": model,
+            "purpose": purpose,
+            "max_tokens": max_tokens,
+            "success": success,
+            "retries": retries,
+        })
+
+    def record_article_stats(self, total, cached, processed):
+        self.articles_total = total
+        self.articles_cached = cached
+        self.articles_processed = processed
+
+    def record_source_result(self, source, success):
+        if success:
+            self.source_success[source] += 1
+        else:
+            self.source_failure[source] += 1
+
+    def report(self):
+        total_elapsed = time.time() - self.start_time
+
+        lines = []
+        lines.append("")
+        lines.append("=" * 64)
+        lines.append("  📊 运行监控报告")
+        lines.append("=" * 64)
+
+        lines.append("")
+        lines.append("── ⏱ 阶段耗时 ──")
+        stage_total = 0.0
+        for key in STAGE_ORDER:
+            if key in self.stage_times:
+                t = self.stage_times[key]
+                stage_total += t
+                label = STAGE_LABELS.get(key, key)
+                lines.append(f"  {label:　<10s}  {t:6.1f}s")
+        lines.append(f"  {'─' * 20}")
+        lines.append(f"  {'合计':　<10s}  {stage_total:6.1f}s / 总运行 {total_elapsed:.0f}s")
+
+        lines.append("")
+        lines.append("── 📡 文章统计 ──")
+        lines.append(f"  总计: {self.articles_total}  缓存命中: {self.articles_cached}  本次处理: {self.articles_processed}")
+        if self.articles_total > 0:
+            cache_rate = self.articles_cached / self.articles_total * 100
+            lines.append(f"  缓存命中率: {cache_rate:.0f}%")
+
+        lines.append("")
+        lines.append("── 🤖 API调用统计 ──")
+        if not self.api_calls:
+            lines.append("  无API调用（全部命中缓存）")
+        else:
+            by_purpose = defaultdict(lambda: {"count": 0, "tokens": 0, "retries": 0, "failed": 0})
+            by_model = defaultdict(lambda: {"count": 0, "tokens": 0})
+            for call in self.api_calls:
+                p = call["purpose"]
+                by_purpose[p]["count"] += 1
+                by_purpose[p]["tokens"] += call["max_tokens"]
+                by_purpose[p]["retries"] += call["retries"]
+                if not call["success"]:
+                    by_purpose[p]["failed"] += 1
+
+                m = call["model"]
+                by_model[m]["count"] += 1
+                by_model[m]["tokens"] += call["max_tokens"]
+
+            lines.append("  [按用途]")
+            for purpose, stats in by_purpose.items():
+                fail_info = f" 失败{stats['failed']}" if stats['failed'] > 0 else ""
+                lines.append(f"    {purpose}: {stats['count']}次 / tokens≈{stats['tokens']} / 重试{stats['retries']}{fail_info}")
+
+            lines.append("  [按模型]")
+            for model, stats in by_model.items():
+                lines.append(f"    {model}: {stats['count']}次 / tokens≈{stats['tokens']}")
+
+        lines.append("")
+        lines.append("── 📰 源抓取统计 ──")
+        total_sources = len(self.source_success) + len(self.source_failure)
+        failed_names = [src for src in self.source_failure if self.source_failure[src] > 0]
+        success_count = total_sources - len(failed_names)
+        lines.append(f"  源总数: {total_sources}  成功: {success_count}  失败: {len(failed_names)}")
+        if failed_names:
+            lines.append(f"  失败源: {', '.join(failed_names)}")
+
+        lines.append("")
+        lines.append("=" * 64)
+
+        report_text = "\n".join(lines)
+        logger.info(report_text)
+        return report_text
+
+
+_monitor = None
+
+
+def get_monitor():
+    global _monitor
+    if _monitor is None:
+        _monitor = RunMonitor()
+    return _monitor
+
+
+def reset_monitor():
+    global _monitor
+    _monitor = RunMonitor()