feat: 基于正则规则引擎的新闻分类 + 发帖自动高亮

classify_rules.json

@@ -0,0 +1,67 @@
+{
+  "_comment": "新闻类型分类规则 - 按优先级从高到低排列，首个匹配的规则生效",
+  "_doc": {
+    "pattern": "正则表达式（不区分大小写）",
+    "exclude": "排除模式列表 - 标题命中任一排除模式则跳过该规则",
+    "require_chinese": "是否也检测中文关键词",
+    "chinese_keywords": "中文关键词列表（任一命中即匹配）"
+  },
+  "rules": [
+    {
+      "type": "java_snapshot",
+      "pattern": "\\b\\d{2}w\\d{2}[a-z]\\b|\\b(?:minecraft|java\\s*edition)[\\s\\S]*?snapshot\\b",
+      "exclude": [],
+      "require_chinese": true,
+      "chinese_keywords": ["快照"]
+    },
+    {
+      "type": "java_prerelease",
+      "pattern": "\\bpre[-\\s]?release\\b",
+      "exclude": [],
+      "require_chinese": true,
+      "chinese_keywords": ["预发布", "预发布版"]
+    },
+    {
+      "type": "java_rc",
+      "pattern": "\\brelease[-\\s]+candidate\\b",
+      "exclude": [],
+      "require_chinese": true,
+      "chinese_keywords": ["候选", "候选版本"]
+    },
+    {
+      "type": "bedrock_beta",
+      "pattern": "\\b(?:beta|preview)\\b[\\s\\S]*?\\d{2,}",
+      "exclude": [],
+      "require_chinese": true,
+      "chinese_keywords": ["预览", "测试版"]
+    },
+    {
+      "type": "bedrock_release",
+      "pattern": "bedrock[\\s\\S]*?(?:edition|changelog)|\\b基岩版\\b",
+      "exclude": ["\\bcaption", "\\bsubtitle", "\\blocali[sz]", "\\btranslation\\b", "\\baccessibility\\b"],
+      "require_chinese": true,
+      "chinese_keywords": ["基岩版"]
+    },
+    {
+      "type": "java_release",
+      "pattern": "java[-\\s]*edition[\\s\\S]*?\\d{2,}|\\d{2,}[\\s\\S]*?java[-\\s]*edition",
+      "exclude": ["\\bsnapshot\\b", "\\bpre[-\\s]?release\\b", "\\brelease[-\\s]+candidate\\b"],
+      "require_chinese": true,
+      "chinese_keywords": ["Java版"]
+    },
+    {
+      "type": "host",
+      "pattern": "\\b(?:xbox|nintendo[-\\s]*switch|playstation|ps[45]\\b|switch\\s*版)",
+      "exclude": [],
+      "require_chinese": false,
+      "chinese_keywords": ["主机"]
+    },
+    {
+      "type": "flash",
+      "pattern": "\\bflash\\b|块讯|速报",
+      "exclude": [],
+      "require_chinese": true,
+      "chinese_keywords": ["块讯", "速报"]
+    }
+  ]
+}

config.json

@@ -106,7 +106,9 @@
         "bedrock_beta": 20,
         "bedrock_release": 20,
         "commentary": 16,
-        "normal": 16
+        "normal": 16,
+        "host": 16,
+        "flash": 16
     },
     "attach_json": true
   }

main.py

@@ -310,6 +310,12 @@ def run_post(processed: list, config: dict, no_image: bool = False, no_json: boo
     success = 0
     failed = 0
 
+    # 处理上次运行中进入审核的待高亮帖子
+    try:
+        poster.process_pending_highlights(save_dir)
+    except Exception as e:
+        print(f"[主] 处理待高亮帖子时出错: {e}")
+
     for stem, txt_path, json_path in processed:
         try:
             print(f"\n[主] 发布: {stem}")
@@ -365,6 +371,12 @@ def run_post_only(config: dict):
         print(f"\n[主] MCBBS 登录失败: {e}")
         return
 
+    # 处理待高亮帖子
+    try:
+        poster.process_pending_highlights(save_dir)
+    except Exception as e:
+        print(f"[主] 处理待高亮帖子时出错: {e}")
+
     for stem, txt_path, json_path in pending:
         try:
             print(f"\n[主] 发布: {stem}")

poster.py

@@ -39,6 +39,23 @@
     'bedrock_release': '基岩版资讯',
     'commentary': '块讯',
     'normal': '块讯',
+    'host': '主机资讯',
+    'flash': '快讯',
+}
+
+# 高亮颜色映射（news_type → Discuz highlight_color 值）
+# 对应 MCBBS 管理操作中的高亮设定
+HIGHLIGHT_COLOR_MAP = {
+    'java_release': 1,      # 正式版 #EE1B2E
+    'bedrock_release': 1,   # 正式版 #EE1B2E
+    'java_snapshot': 4,     # 快照/预览版 #3C9D40
+    'java_prerelease': 4,   # 预发布版 #3C9D40
+    'java_rc': 4,           # 候选版本 #3C9D40
+    'bedrock_beta': 4,      # 测试版 #3C9D40
+    'flash': 8,             # 块讯 #EC1282
+    'host': 2,              # 主机资讯 #EE5023
+    'peripheral': 6,        # 周边消息 #2B65B7
+    'normal': 8,            # 块讯（兜底）#EC1282
 }
 
 UA = (
@@ -562,6 +579,192 @@ def post_thread(self, title: str, message: str,
 
         raise RuntimeError(f"发帖结果不明: {r.url}")
 
+    def _apply_highlight(self, thread_url: str, highlight_color: int) -> bool:
+        """对已发帖子应用高亮（Discuz topicadmin 管理操作）
+
+        需要版主/管理员权限。通过 topicadmin moderate 端点提交高亮请求。
+        """
+        if not self.session:
+            return False
+        if highlight_color <= 0:
+            return False
+
+        tid_match = re.search(r'thread-(\d+)-', thread_url)
+        if not tid_match:
+            print("    ⚠ 无法提取帖子 ID，跳过高亮")
+            return False
+        tid = tid_match.group(1)
+
+        try:
+            # 确保 formhash 有效
+            if not self.formhash:
+                r = self.session.get(f"{self.base_url}/forum.php")
+                r.raise_for_status()
+                self.formhash = extract_formhash(r.text)
+
+            # 通过 topicadmin 获取高亮表单
+            moderate_url = (
+                f"{self.base_url}/forum.php?mod=topicadmin&action=moderate"
+                f"&fid={self.forum_fid}&tid={tid}&optgroup=1&infloat=yes&inajax=1"
+            )
+            referer = f"{self.base_url}/thread-{tid}-1-1.html"
+            self.session.headers.update({"Referer": referer, "X-Requested-With": "XMLHttpRequest"})
+
+            r_get = self.session.get(moderate_url)
+            self.session.headers.pop("X-Requested-With", None)
+
+            # 检查是否有权限
+            if "没有权限" in r_get.text or "alert_error" in r_get.text:
+                err_m = re.search(r'class="alert_error">([^<]+)', r_get.text)
+                err_msg = err_m.group(1).strip() if err_m else "无管理权限"
+                print(f"    ⚠ 高亮失败: {err_msg}")
+                self.session.headers.pop("Referer", None)
+                return False
+
+            # 从返回的表单中提取 formhash
+            form_fh = re.search(r'name="formhash"\s+value="([a-f0-9]+)"', r_get.text)
+            if not form_fh:
+                print("    ⚠ 无法提取高亮表单 formhash")
+                self.session.headers.pop("Referer", None)
+                return False
+
+            # 提交高亮（Discuz topicadmin 标准参数）
+            action_url = (
+                f"{self.base_url}/forum.php?mod=topicadmin&action=moderate"
+                f"&optgroup=1&modsubmit=yes&infloat=yes"
+            )
+            post_data = {
+                "formhash": form_fh.group(1),
+                "fid": str(self.forum_fid),
+                "redirect": f"{self.base_url}/thread-{tid}-1-1.html",
+                "handlekey": "mods",
+                "moderate[]": tid,
+                "operations[]": "highlight",
+                "highlight_color": str(highlight_color),
+                "highlight_style[1]": "0",
+                "highlight_style[2]": "0",
+                "highlight_style[3]": "0",
+                "highlight_bgcolor": "",
+                "expirationhighlight": "",
+                "sendreasonpm": "on",
+            }
+
+            r_post = self.session.post(action_url, data=post_data)
+            self.session.headers.pop("Referer", None)
+
+            # 检查结果
+            resp = r_post.text
+            if "succeedhandle" in resp or "成功" in resp:
+                return True
+            if "没有权限" in resp:
+                print("    ⚠ 高亮失败: 账号无管理权限")
+                return False
+            if "errorhandle" in resp:
+                err_m = re.search(r"errorhandle[^']*'([^']+)'", resp)
+                print(f"    ⚠ 高亮失败: {err_m.group(1) if err_m else '未知错误'}")
+                return False
+            return False
+        except Exception as e:
+            print(f"    ⚠ 高亮失败: {e}")
+            return False
+
+    def _save_pending_highlight(self, post_url: str, original_title: str,
+                                translated_title: str, news_dir: str):
+        """保存待高亮的帖子信息到 JSON 文件，等待审核通过后处理"""
+        from utils import classify_article_type
+        news_type = None
+        if original_title:
+            news_type = classify_article_type(original_title, chinese=False, fallback=None)
+        if not news_type:
+            news_type = classify_article_type(translated_title, chinese=True, fallback=None)
+        if not news_type:
+            news_type = "normal"
+        highlight_color = HIGHLIGHT_COLOR_MAP.get(news_type, 0)
+
+        pending_file = os.path.join(news_dir, ".pending_highlights.json")
+        pending = []
+        if os.path.exists(pending_file):
+            try:
+                with open(pending_file, encoding="utf-8") as f:
+                    pending = json.load(f)
+            except (OSError, json.JSONDecodeError):
+                pass
+
+        # 提取 thread ID
+        tid_match = re.search(r'thread-(\d+)-', post_url)
+        tid = tid_match.group(1) if tid_match else ""
+
+        pending.append({
+            "url": post_url,
+            "tid": tid,
+            "title": original_title,
+            "translated_title": translated_title,
+            "news_type": news_type,
+            "highlight_color": highlight_color,
+            "created_at": time.strftime("%Y-%m-%dT%H:%M:%S"),
+        })
+
+        try:
+            with open(pending_file, "w", encoding="utf-8") as f:
+                json.dump(pending, f, ensure_ascii=False, indent=2)
+            print(f"    📝 已保存待高亮信息（共{len(pending)}条待处理）")
+        except OSError as e:
+            print(f"    ⚠ 保存待高亮信息失败: {e}")
+
+    def process_pending_highlights(self, news_dir: str) -> int:
+        """处理待高亮的帖子（审核通过后调用）。返回成功数。"""
+        pending_file = os.path.join(news_dir, ".pending_highlights.json")
+        if not os.path.exists(pending_file):
+            return 0
+
+        try:
+            with open(pending_file, encoding="utf-8") as f:
+                pending = json.load(f)
+        except (OSError, json.JSONDecodeError):
+            return 0
+
+        if not pending:
+            return 0
+
+        print(f"\n[*] 处理 {len(pending)} 条待高亮帖子...")
+        remaining = []
+        success = 0
+        color_names = {
+            1: "红色(正式版)", 2: "橙色(主机)",
+            4: "绿色(快照/测试版)", 6: "蓝色(周边)",
+            8: "粉色(块讯)",
+        }
+
+        for item in pending:
+            url = item.get("url", "")
+            color = item.get("highlight_color", 0)
+            title = item.get("translated_title") or item.get("title", "")
+
+            if not url.startswith("http") or not color:
+                continue
+
+            print(f"  尝试高亮: {title[:50]}...")
+            if self._apply_highlight(url, color):
+                print(f"    ✓ 高亮成功: {color_names.get(color, str(color))}")
+                success += 1
+            else:
+                # 可能仍未审核通过，保留待处理
+                remaining.append(item)
+
+        # 更新待处理文件
+        try:
+            if remaining:
+                with open(pending_file, "w", encoding="utf-8") as f:
+                    json.dump(remaining, f, ensure_ascii=False, indent=2)
+                print(f"  剩余 {len(remaining)} 条待处理")
+            else:
+                os.remove(pending_file)
+                print("  所有待高亮帖子已处理完毕")
+        except OSError:
+            pass
+
+        return success
+
     def post_news_file(self, stem: str, txt_path: str, json_path: str,
                        news_dir: str, no_image: bool = False,
                        attach_json: bool = True) -> str:
@@ -627,7 +830,39 @@ def post_news_file(self, stem: str, txt_path: str, json_path: str,
         else:
             print(f"    ⚠ 警告: 未找到分类 {module_type} 的 sortid 配置")
 
-        return self.post_thread(title, message, attachment_ids=attachment_ids, sortid=sortid)
+        post_url = self.post_thread(title, message, attachment_ids=attachment_ids, sortid=sortid)
+
+        # 发帖后自动高亮（审核中的帖子跳过，记录待处理）
+        need_moderation = "(需审核)" in (post_url or "")
+        if post_url and post_url.startswith("http") and not need_moderation:
+            # 用原标题检测新闻类型用于高亮（英文优先，中文补充）
+            from utils import classify_article_type
+            news_type = None
+            if original_title:
+                news_type = classify_article_type(original_title, chinese=False, fallback=None)
+            if not news_type:
+                news_type = classify_article_type(title, chinese=True, fallback=None)
+            # 兜底：无法识别的新闻默认归为“块讯”，确保所有帖子都有高亮
+            if not news_type:
+                news_type = "normal"
+            highlight_color = HIGHLIGHT_COLOR_MAP.get(news_type, 0)
+            if highlight_color:
+                color_names = {
+                    1: "红色(正式版)", 2: "橙色(主机)",
+                    4: "绿色(快照/测试版)", 6: "蓝色(周边)",
+                    8: "粉色(块讯)",
+                }
+                print(f"    高亮: {color_names.get(highlight_color, str(highlight_color))}")
+                if self._apply_highlight(post_url, highlight_color):
+                    print("    ✓ 高亮设置成功！")
+                else:
+                    print("    ⚠ 高亮设置失败（不影响发帖）")
+        elif need_moderation:
+            # 审核中的帖子：记录待高亮信息，下次运行时补上
+            print("    ⏳ 帖子进入审核，高亮待审核通过后处理")
+            self._save_pending_highlight(post_url, original_title, title, news_dir)
+
+        return post_url
 
 
 # ── 状态管理 ─────────────────────────────────────────