From c644786e8783be88b08f27662e501ed86072687a Mon Sep 17 00:00:00 2001 From: Kubbo <390378816@qq.com> Date: Sat, 3 Aug 2024 02:27:37 +0800 Subject: [PATCH] =?UTF-8?q?=E4=BF=AE=E6=94=B9=E5=AE=8C=E6=AF=95?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- README.md | 6 +++--- markdown.py | 62 ++++++++++++++++------------------------------------- 2 files changed, 22 insertions(+), 46 deletions(-) diff --git a/README.md b/README.md index cc8f97f..93b1645 100644 --- a/README.md +++ b/README.md @@ -1,14 +1,14 @@ 一款用来适配Dify的chatbot的返回的markdown格式返回的消息处理 + 提取Dify返回的Markdown图片链接中的网址，并修改ReplyType为IMAGE_URL，以便DoW自动下载Markdown链接中的图片； -+ 提取Dify返回的包含 https://s.coze.cn/t/xxx 网址的Markdown链接中的图片网址，并修改ReplyType为IMAGE_URL，以便CoW自动下载Markdown链接中的图片； ++ 提取Dify返回的包含网址的Markdown链接中的图片网址，并修改ReplyType为IMAGE_URL，以便CoW自动下载Markdown链接中的图片； + 去掉每行结尾的Markdown链接中网址部分的小括号，避免微信误以为“)”是网址的一部分导致微信中无法打开该页面。 **安装方法：** ```sh -#installp https://github.com/wangxyd/nicecoze.git +#installp https://github.com/AceyKubbo/dow-markdown.git #scanp ``` @@ -18,5 +18,5 @@ **更新方法：** ```sh -#updatep nicecoze +#updatep dow_markdown ``` \ No newline at end of file diff --git a/markdown.py b/markdown.py index 56682e5..0c5d456 100644 --- a/markdown.py +++ b/markdown.py @@ -2,6 +2,7 @@ import re import requests +import os import plugins from bridge.reply import Reply, ReplyType @@ -10,23 +11,22 @@ from plugins import * @plugins.register( - name="dow-markdown", + name="dow_markdown", desire_priority=66, hidden=False, desc="优化markdown返回结果中的图片和网址链接。", version="0.1", author="Kubbo", ) -class NiceCoze(Plugin): +class dow_markdown(Plugin): def __init__(self): super().__init__() try: - self.handlers[Event.ON_DECORATE_REPLY] = self.on_decorate_reply - logger.info("[Nicecoze] inited.") + self.handlers[Event.ON_HANDLE_CONTEXT] = self.on_decorate_reply + logger.info("[dow_markdown] inited.") except Exception as e: - logger.warn("[Nicecoze] init failed, ignore.") + logger.warn("[dow_markdown] init failed, ignore.") raise e - def on_decorate_reply(self, e_context: EventContext): if e_context["reply"].type != ReplyType.TEXT: return @@ -37,46 +37,22 @@ class NiceCoze(Plugin): # 避免图片无法下载时，重复调用插件导致没有响应的问题 if content.startswith("[DOWNLOAD_ERROR]"): return - # 提取Coze返回的Markdown图片链接中的网址，并修改ReplyType为IMAGE_URL，以便CoW自动下载Markdown链接中的图片 - #if all(x in content for x in ['![', 'http']) and any(x in content for x in ['jpg', 'jpeg', 'png', 'gif', 'bmp', 'webp']): - if 'http' in content and any(x in content for x in ['.jpg', '.jpeg', '.png', '.gif', '.bmp', '.webp']): - logger.debug(f"[Nicecoze] starting decorate_markdown_image, content={content}") - replies = self.decorate_markdown_image(content) - if replies: - logger.info(f"[Nicecoze] sending {len(replies)} images ...") - e_context["reply"].content = "[DOWNLOAD_ERROR]\n" + e_context["reply"].content - for reply in replies: - channel.send(reply, context) - #e_context["reply"] = Reply(ReplyType.TEXT, f"{len(replies)}张图片已发送，收到了吗？") - # “x张图片已发送，收到了吗？”提示的初衷是告诉我们画/搜了几张图片以及下载/发送失败了几张图片，可以将e_context["reply"]设置为None关闭该提示！ - e_context["reply"] = None - e_context.action = EventAction.BREAK_PASS - return - # 提取Coze返回的包含https://s.coze.cn/t/xxx网址的Markdown链接中的图片网址 - markdown_s_coze_cn = r"([\S\s]*)\!?\[(?P.*)\]\((?Phttps\:\/\/s\.coze\.cn\/t\/[\S]*?)\)([\S\s]*)" - match_obj_s_coze_cn = re.fullmatch(markdown_s_coze_cn, content) - if match_obj_s_coze_cn and match_obj_s_coze_cn.group('link_url'): - link_url = match_obj_s_coze_cn.group('link_url') - logger.info(f"[Nicecoze] match_obj_s_coze_cn found, link_url={link_url}") - response = requests.get(url=link_url, allow_redirects=False) - original_url = response.headers.get('Location') - if response.status_code in [301, 302] and original_url and any(x in original_url for x in ['.jpg', '.jpeg', '.png', '.gif', '.bmp', '.webp']): - logger.info(f"[Nicecoze] match_obj_s_coze_cn found and original_url is a image url, original_url={original_url}") - reply = Reply(ReplyType.IMAGE_URL, original_url) - e_context["reply"] = reply - e_context.action = EventAction.BREAK_PASS - return - else: - logger.info(f"[Nicecoze] match_obj_s_coze_cn found but failed to get original_url or original_url is not a image url, response.status_code={response.status_code}, original_url={original_url}") + if bool(re.search(r'\!\[[^\]]+\]\(?',content)): + host = os.environ.get('DIFY_API_BASE', 'http://121.37.155.68:35801') + image_path = re.search(r'!\[.*\]\((.*?)\)',reply.content).group(1) + logger.info(f"提取到的数据==>host:{host},url:{image_path}") + reply = Reply(ReplyType.IMAGE_URL, f"{host}{image_path}") + e_context["reply"] = reply + e_context.action = EventAction.BREAK_PASS # 去掉每行结尾的Markdown链接中网址部分的小括号，避免微信误以为“)”是网址的一部分导致微信中无法打开该页面 content_list = content.split('\n') new_content_list = [re.sub(r'\((https?://[^\s]+)\)$', r' \1', line) for line in content_list] if new_content_list != content_list: - logger.info(f"[Nicecoze] parenthesis in the url has been removed, content={content}") + logger.info(f"[dow_markdown] parenthesis in the url has been removed, content={content}") reply = Reply(ReplyType.TEXT, '\n'.join(new_content_list).strip()) e_context["reply"] = reply except Exception as e: - logger.warn(f"[Nicecoze] on_decorate_reply failed, content={content}, error={e}") + logger.warn(f"[dow_markdown] on_decorate_reply failed, content={content}, error={e}") finally: e_context.action = EventAction.CONTINUE @@ -86,7 +62,7 @@ class NiceCoze(Plugin): match_obj_official = re.fullmatch(markdown_image_official, content) if match_obj_official and match_obj_official.group('image_url'): image_name, image_url = match_obj_official.group('image_name'), match_obj_official.group('image_url') - logger.info(f"[Nicecoze] markdown_image_official found, image_name={image_name}, image_url={image_url}") + logger.info(f"[dow_markdown] markdown_image_official found, image_name={image_name}, image_url={image_url}") reply = Reply(ReplyType.IMAGE_URL, image_url) return [reply] # 完全匹配一张Markdown图片（格式：`![name](url)`） @@ -94,7 +70,7 @@ class NiceCoze(Plugin): match_obj_single = re.fullmatch(markdown_image_single, content, re.DOTALL) if match_obj_single and match_obj_single.group('image_url'): image_name, image_url = match_obj_single.group('image_name'), match_obj_single.group('image_url') - logger.info(f"[Nicecoze] markdown_image_single found, image_name={image_name}, image_url={image_url}") + logger.info(f"[dow_markdown] markdown_image_single found, image_name={image_name}, image_url={image_url}") reply = Reply(ReplyType.IMAGE_URL, image_url) return [reply] # 匹配多张Markdown图片(格式：`url\n![Image](url)`) @@ -103,13 +79,13 @@ class NiceCoze(Plugin): replies = [] for match in match_iter_multi: image_url = match.group('image_url') - logger.info(f"[Nicecoze] markdown_image_multi found, image_url={image_url}") + logger.info(f"[dow_markdown] markdown_image_multi found, image_url={image_url}") reply = Reply(ReplyType.IMAGE_URL, image_url) replies.append(reply) if replies: return replies if content.startswith('![') and 'http' in content and any(img in content for img in ['jpg', 'jpeg', 'png', 'gif', 'bmp', 'webp']): - logger.info(f"[Nicecoze] it seems markdown image in the content but not matched, content={content}.") + logger.info(f"[dow_markdown] it seems markdown image in the content but not matched, content={content}.") def get_help_text(self, **kwargs): return "优化Coze返回结果中的图片和网址链接。"