From 60b372cd88b0836f8c359179fe3f6cd2cda5de43 Mon Sep 17 00:00:00 2001 From: liuwei Date: Thu, 23 Apr 2026 11:40:11 +0800 Subject: [PATCH] =?UTF-8?q?=E9=85=8D=E7=BD=AE=E6=A8=A1=E6=9D=BF=E6=88=AA?= =?UTF-8?q?=E5=9B=BE=E8=A7=86=E5=8F=A3=E5=8F=82=E6=95=B0=E5=B9=B6=E8=BE=93?= =?UTF-8?q?=E5=87=BA=E7=94=9F=E6=95=88=E5=80=BC=E6=97=A5=E5=BF=97?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- plugins/message_summary/config.toml | 8 ++++++++ plugins/message_summary/main.py | 7 +++++++ 2 files changed, 15 insertions(+) diff --git a/plugins/message_summary/config.toml b/plugins/message_summary/config.toml index f584b39..8544a9b 100644 --- a/plugins/message_summary/config.toml +++ b/plugins/message_summary/config.toml @@ -18,3 +18,11 @@ image_format = "png" summary_image_mode = "template" # 总结卡片模板路径（相对项目根目录） summary_image_template_path = "plugins/message_summary/templates/gemini_summary_card.html" +# 模板模式截图视口配置（仅 template 模式生效） +# 说明： +# 1. 这里可以覆盖 md2image 默认 780 宽； +# 2. 模板卡片宽度约 420，截图建议 460~520 区间； +# 3. scale 提高后字体与线条会更清晰。 +template_viewport_width = 500 +template_viewport_height = 960 +template_device_scale_factor = 2.0 diff --git a/plugins/message_summary/main.py b/plugins/message_summary/main.py index df7b485..913ebf4 100644 --- a/plugins/message_summary/main.py +++ b/plugins/message_summary/main.py @@ -117,6 +117,13 @@ class MessageSummaryPlugin(MessagePluginInterface): self._template_viewport_width = int(output_config.get("template_viewport_width", 780)) self._template_viewport_height = int(output_config.get("template_viewport_height", 960)) self._template_device_scale_factor = float(output_config.get("template_device_scale_factor", 1.2)) + # 启动时输出模板截图视口参数，便于排查“配置未生效仍是 780 宽”的问题。 + self.LOG.info( + f"[{self.name}] 模板截图视口: " + f"width={self._template_viewport_width}, " + f"height={self._template_viewport_height}, " + f"scale={self._template_device_scale_factor}" + ) self.llm_client = UnifiedLLMClient(api_config) self._api_mode = self.llm_client.mode or self._api_mode self._response_mode = self.llm_client.response_mode or self._response_mode