update manuel

client/.dockerignore

@@ -11,4 +11,4 @@ docker-compose.yaml
 Dockerfile
 README.md
 backend/__pycache__
-backend/AWtest
+backend/WStest

client/.gitignore

@@ -1,4 +1,4 @@
 .env
 .venv/
 pb/pb_data/
-backend/AWtest/
+backend/WStest/

client/README.md

@@ -18,16 +18,21 @@
 - character 以什么身份挖掘线索（这决定了llm的关注点和立场）
 - focus 关注什么方面的线索
 - focus_type 线索类型
-- good_samples 你希望llm给出的线索描述模式（给两个sample）
+- good_samples1 你希望llm给出的线索描述模式（给两个sample）
+- good_samples2 你希望llm给出的线索描述模式（给两个sample）
 - bad_samples 规避的线索描述模式
 - report_type 报告类型
 
-- 【sites] 大类下面列出你的信源。一行一个网址。
+### 4、编辑 sites.txt 文件
+
+这个文件指定了需要本地执行的监控的信源，一行一个网址，支持随时更改，每次执行任务前会读取最新的。
 
 如果你只爬取配置了专有爬虫的信源的话，可以直接编辑scrapers/__init__.py 中的scraper_map，这里都留空就好
 
 专有爬虫的说明见 backend/scrapers/README.md
 
+**注：虽然wiseflow client配置了通用爬虫，对于新闻类静态网页有一定的爬取和解析效果，但我们还是强烈建议使用我们的数据订阅服务或者自写专业爬虫。**
+
 ## 参考：各服务注册地址
 
 - 阿里灵积大模型接口：https://dashscope.aliyun.com/

client/backend/background_task.py

@@ -4,22 +4,13 @@
 import schedule
 import time
 from work_process import ServiceProcesser
-import configparser
-
-config = configparser.ConfigParser()
-config.read('../config.ini')
-
-if config.has_section('sites'):
-    web_pages = config['sites']
-    urls = [value for key, value in web_pages.items()]
-else:
-    urls = []
 
 sp = ServiceProcesser()
-sp(sites=urls)
 
-'''
+
 def task():
+    with open('../sites.txt', 'r', encoding='utf-8') as f:
+        urls = [line.strip() for line in f.readlines() if line.strip()]
     sp(sites=urls)
 
 
@@ -29,6 +20,3 @@ schedule.every().day.at("01:17").do(task)
 while True:
     schedule.run_pending()
     time.sleep(60)
-site1 = https://www.welivesecurity.com/en/
-site2 = https://www.scmagazine.com/
-'''

client/backend/general_utils.py

@@ -125,6 +125,25 @@ def is_chinese(string):
     return (non_chinese_count/len(string)) < 0.68
 
 
+def extract_and_convert_dates(input_string):
+    # 定义匹配不同日期格式的正则表达式
+    patterns = [
+        r'(\d{4})-(\d{2})-(\d{2})',  # 匹配YYYY-MM-DD格式
+        r'(\d{4})/(\d{2})/(\d{2})',  # 匹配YYYY/MM/DD格式
+        r'(\d{4})\.(\d{2})\.(\d{2})',  # 匹配YYYY.MM.DD格式
+        r'(\d{4})\\(\d{2})\\(\d{2})',  # 匹配YYYY\MM\DD格式
+        r'(\d{4})(\d{2})(\d{2})'  # 匹配YYYYMMDD格式
+    ]
+
+    matches = []
+    for pattern in patterns:
+        matches = re.findall(pattern, input_string)
+        if matches:
+            break
+    if matches:
+        return ''.join(matches[0])
+    return None
+
 """
 # from InternLM/huixiangdou 
 # another awsome work

client/backend/get_search.py

@@ -85,7 +85,7 @@ def search_insight(keyword: str, exist_urls: list[Union[str, Path]], knowledge:
         if url in exist_urls:
             continue
         exist_urls.append(url)
-        flag, value = simple_crawler(url, logger)
+        flag, value = simple_crawler(url)
         if flag != 11:
             continue
         from_site = urlparse(url).netloc

client/backend/scrapers/README.md

@@ -15,7 +15,6 @@
 输入：
 - expiration： datetime的date.date()对象，爬虫应该只抓取这之后（含这一天）的文章
 - existings：[str], 数据库已有文章的url列表，爬虫应该忽略这个列表里面的url
-- logger：主流程的logger对象，如果爬虫需要单独logger，这个logger接收了可以不用
 
 输出：
 - [dict]，返回结果列表，每个dict代表一个文章，格式如下：

client/backend/scrapers/general_scraper.py

@@ -2,18 +2,22 @@ from pathlib import Path
 from urllib.parse import urlparse
 import re
 from .simple_crawler import simple_crawler
-import json
+import httpx
-import requests
 from bs4 import BeautifulSoup
 from bs4.element import Comment
 from llms.dashscope_wrapper import dashscope_llm
 from datetime import datetime, date
 from requests.compat import urljoin
 import chardet
+from general_utils import extract_and_convert_dates
+from get_logger import get_logger
+import os
 
 
 header = {
         'User-Agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_14_6) AppleWebKit/605.1.15 (KHTML, like Gecko) Chrome/112.0.0.0 Safari/604.1 Edg/112.0.100.0'}
+project_dir = os.environ.get("PROJECT_DIR", "")
+logger = get_logger(name='general_scraper', file=os.path.join(project_dir, f'general_scraper.log'))
 
 
 def tag_visible(element: Comment) -> bool:
@@ -35,83 +39,56 @@ def text_from_soup(soup: BeautifulSoup) -> str:
 
 
 def parse_html_content(out: str) -> dict:
-    # 发现llm出来的结果有时会在键值或者内容的引号外面出现\n \t安全起见全部去除，反正后续分析时llm也不看内容的换行这些
+    dct = {'title': '', 'abstract': '', 'content': '', 'publish_time': ''}
     pattern = re.compile(r'\"\"\"(.*?)\"\"\"', re.DOTALL)
     result = pattern.findall(out)
-    out = result[0]
+    result = result[0].strip()
-    dict_str = out.strip("```").strip("python").strip("json").strip()
+    dict_strs = result.split('||')
-    dict_str = dict_str.replace("\n", "").replace("\t", "")
+    if not dict_strs:
-    # 先正则解析出{}中的内容
+        dict_strs = result.split('|||')
-    dict_str = re.findall(r'{(.*?)}', dict_str)
+        if not dict_strs:
-    # dict_str = dict_str[0].replace("'", '"') #会误伤
+            return dct
-    # json loads 要求双引号, 且需要把\n等转译
+    if len(dict_strs) == 3:
-    dct = json.loads('{' + dict_str[0] + '}')
+        dct['title'] = dict_strs[0].strip()
-    date_str = re.findall(r"\d{4}-\d{2}-\d{2}", dct['publish_time'])
+        dct['content'] = dict_strs[1].strip()
-    if date_str:
+    elif len(dict_strs) == 4:
-        dct['publish_time'] = date_str[0].replace("-", "")
+        dct['title'] = dict_strs[0].strip()
+        dct['content'] = dict_strs[2].strip()
+        dct['abstract'] = dict_strs[1].strip()
     else:
-        date_str = re.findall(r"\d{4}\.\d{2}\.\d{2}", dct['publish_time'])
+        return dct
+    date_str = extract_and_convert_dates(dict_strs[-1])
     if date_str:
-            dct['publish_time'] = date_str[0].replace(".", "")
+        dct['publish_time'] = date_str
-        else:
-            date_str = re.findall(r"\d{4}\d{2}\d{2}", dct['publish_time'])
-            if date_str:
-                dct['publish_time'] = date_str[0]
     else:
         dct['publish_time'] = datetime.strftime(datetime.today(), "%Y%m%d")
     return dct
 
 
-sys_info = '''你是一个html网页解析器，你将接收一段用户从网页html文件中提取的文本，请解析出其标题、摘要、内容和发布日期。
+# qwen1.5-72b解析json格式太容易出错，网页上的情况太多，比如经常直接使用英文的"，这样后面json.loads就容易出错……
-发布日期的格式为：XXXX-XX-XX，如果找不到则为空。内容不要包含标题、作者和发布日期。
+sys_info = '''你是一个html网页解析器，你将接收一段用户从网页html文件中提取的文本，请解析出其标题、摘要、内容和发布日期，发布日期格式为YYYY-MM-DD。
-请务必按照Python字典的格式输出，key和value使用双引号包裹，key分别为：title、abstract、content和publish_time。输出结果请整体用三引号包裹，如下所示：
+结果请按照以下格式返回（整体用三引号包裹）：
 """
-{"title": "解析出的标题", "abstract": "解析出的摘要", "content": "解析出的内容", "publish_time": "解析出的发布日期XXXX-XX-XX"}
+标题||摘要||内容||发布日期XXXX-XX-XX
-"""'''
+"""
+'''
 
 
-def llm_crawler(url: str | Path, logger=None) -> (int, dict):
+def llm_crawler(url: str | Path) -> (int, dict):
     """
     返回文章信息dict和flag，负数为报错，0为没有结果，11为成功
     参考：https://mp.weixin.qq.com/s/4J-kofsfFDiV1FxGlTJLfA
-    测试URL：
-    url = "https://so.html5.qq.com/page/real/search_news?docid=70000021_40665eb6afe80152"
-    url = "https://mp.weixin.qq.com/s?src=11&timestamp=1709999167&ver=5128&signature=e0Tssc4COc*p-RkKaPwUMrGePUxko8N621VxARnI8uKDg*l5C7Z8gBC6RDUAnyGqvmzJ5WEzvaO-T7GvMRw9LwNaJS3Hh2tyaITdmsaVtY9JsSmsidX6u4SqxirGsRdo&new=1"
     """
     # 发送 HTTP 请求获取网页内容
     try:
-        response = requests.get(url, timeout=60)
+        with httpx.Client() as client:
-    except:
+            response = client.get(url, headers=header, timeout=30)
-        if logger:
-            logger.error(f"cannot connect {url}")
-        else:
-            print(f"cannot connect {url}")
-        return -7, {}
-
-    if response.status_code != 200:
-        if logger:
-            logger.error(f"cannot connect {url}")
-        else:
-            print(f"cannot connect {url}")
-        return -7, {}
-
             rawdata = response.content
             encoding = chardet.detect(rawdata)['encoding']
-    if encoding is not None and encoding.lower() == 'utf-8':
-        try:
             text = rawdata.decode(encoding)
-        except:
+    except Exception as e:
-            if logger:
+        logger.error(e)
-                logger.error(f"{url} decode error, aborting")
+        return -7, {}
-            else:
-                print(f"{url} decode error, aborting")
-            return 0, {}
-    else:
-        if logger:
-            logger.error(f"{url} undetected coding, aborting")
-        else:
-            print(f"{url} undetected coding, aborting")
-        return 0, {}
 
     # 使用 BeautifulSoup 解析 HTML 内容
     soup = BeautifulSoup(text, "html.parser")
@@ -120,17 +97,12 @@ def llm_crawler(url: str | Path, logger=None) -> (int, dict):
     html_lines = [line.strip() for line in html_lines if line.strip()]
     html_text = "\n".join(html_lines)
     if len(html_text) > 29999:
-        if logger:
         logger.warning(f"{url} content too long for llm parsing")
-        else:
-            print(f"{url} content too long for llm parsing")
         return 0, {}
 
-    if not html_text or html_text.startswith('服务器错误') or html_text.startswith('您访问的页面') or html_text.startswith('403'):
+    if not html_text or html_text.startswith('服务器错误') or html_text.startswith('您访问的页面') or html_text.startswith('403')\
-        if logger:
+            or html_text.startswith('出错了'):
         logger.warning(f"can not get {url} from the Internet")
-        else:
-            print(f"can not get {url} from the Internet")
         return -7, {}
 
     messages = [
@@ -140,19 +112,13 @@ def llm_crawler(url: str | Path, logger=None) -> (int, dict):
     llm_output = dashscope_llm(messages, "qwen1.5-72b-chat", logger=logger)
     try:
         info = parse_html_content(llm_output)
-    except Exception as e:
+    except Exception:
         msg = f"can not parse {llm_output}"
-        if logger:
+        logger.debug(msg)
-            logger.warning(msg)
-        else:
-            print(msg)
         return 0, {}
 
-    if len(info['title']) < 5 or len(info['content']) < 24:
+    if len(info['title']) < 4 or len(info['content']) < 24:
-        if logger:
+        logger.debug(f"{info} not valid")
-            logger.warning(f"{info} not valid")
-        else:
-            print(f"{info} not valid")
         return 0, {}
 
     info["url"] = str(url)
@@ -185,21 +151,12 @@ def llm_crawler(url: str | Path, logger=None) -> (int, dict):
     return 11, info
 
 
-def general_scraper(site: str, expiration: date, existing: list[str], logger=None) -> list[dict]:
+def general_scraper(site: str, expiration: date, existing: list[str]) -> list[dict]:
     try:
-        response = requests.get(site, header, timeout=60)
+        with httpx.Client() as client:
-    except:
+            response = client.get(site, headers=header, timeout=30)
-        if logger:
+    except Exception as e:
-            logger.error(f"cannot connect {site}")
+        logger.error(e)
-        else:
-            print(f"cannot connect {site}")
-        return []
-
-    if response.status_code != 200:
-        if logger:
-            logger.error(f"cannot connect {site}")
-        else:
-            print(f"cannot connect {site}")
         return []
 
     page_source = response.text
@@ -209,25 +166,18 @@ def general_scraper(site: str, expiration: date, existing: list[str], logger=Non
     base_url = parsed_url.scheme + '://' + parsed_url.netloc
     urls = [urljoin(base_url, link["href"]) for link in soup.find_all("a", href=True)]
     if not urls:
-        if logger:
         logger.warning(f"can not find any link from {site}, maybe it's an article site...")
         if site in existing:
-            if logger:
+            logger.debug(f"{site} has been crawled before, skip it")
-                logger.warning(f"{site} has been crawled before, skip it")
-            else:
-                print(f"{site} has been crawled before, skip it")
             return []
-        flag, result = simple_crawler(site, logger=logger)
+        flag, result = simple_crawler(site)
         if flag != 11:
-            flag, result = llm_crawler(site, logger=logger)
+            flag, result = llm_crawler(site)
             if flag != 11:
                 return []
         publish_date = datetime.strptime(result['publish_time'], '%Y%m%d')
         if publish_date.date() < expiration:
-            if logger:
+            logger.debug(f"{site} is too old, skip it")
-                logger.warning(f"{site} is too old, skip it")
-            else:
-                print(f"{site} is too old, skip it")
             return []
         else:
             return [result]
@@ -235,23 +185,17 @@ def general_scraper(site: str, expiration: date, existing: list[str], logger=Non
     articles = []
     for url in urls:
         if url in existing:
-            if logger:
+            logger.debug(f"{url} has been crawled before, skip it")
-                logger.warning(f"{url} has been crawled before, skip it")
-            else:
-                print(f"{url} has been crawled before, skip it")
             continue
         existing.append(url)
-        flag, result = simple_crawler(url, logger=logger)
+        flag, result = simple_crawler(url)
         if flag != 11:
-            flag, result = llm_crawler(url, logger=logger)
+            flag, result = llm_crawler(url)
             if flag != 11:
                 continue
         publish_date = datetime.strptime(result['publish_time'], '%Y%m%d')
         if publish_date.date() < expiration:
-            if logger:
+            logger.debug(f"{url} is too old, skip it")
-                logger.warning(f"{url} is too old, skip it")
-            else:
-                print(f"{url} is too old, skip it")
         else:
             articles.append(result)
 

client/backend/scrapers/simple_crawler.py

@@ -1,88 +1,52 @@
 from gne import GeneralNewsExtractor
-import requests
+import httpx
 from bs4 import BeautifulSoup
 from datetime import datetime
 from pathlib import Path
-import re
+from general_utils import extract_and_convert_dates
 import chardet
-
+from get_logger import get_logger
+import os
 
 extractor = GeneralNewsExtractor()
 header = {
         'User-Agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_14_6) AppleWebKit/605.1.15 (KHTML, like Gecko) Chrome/112.0.0.0 Safari/604.1 Edg/112.0.100.0'}
 
+project_dir = os.environ.get("PROJECT_DIR", "")
+logger = get_logger(name='simple_crawler', file=os.path.join(project_dir, f'simple_crawler.log'))
 
-def simple_crawler(url: str | Path, logger=None) -> (int, dict):
+
+def simple_crawler(url: str | Path) -> (int, dict):
     """
     返回文章信息dict和flag，负数为报错，0为没有结果，11为成功
     """
     try:
-        response = requests.get(url, header, timeout=60)
+        with httpx.Client() as client:
-    except:
+            response = client.get(url, headers=header, timeout=30)
-        if logger:
-            logger.error(f"cannot connect {url}")
-        else:
-            print(f"cannot connect {url}")
-        return -7, {}
-
-    if response.status_code != 200:
-        if logger:
-            logger.error(f"cannot connect {url}")
-        else:
-            print(f"cannot connect {url}")
-        return -7, {}
-
             rawdata = response.content
             encoding = chardet.detect(rawdata)['encoding']
-    if encoding is not None and encoding.lower() == 'utf-8':
-        try:
             text = rawdata.decode(encoding)
-        except:
+    except Exception as e:
-            if logger:
+        logger.error(e)
-                logger.error(f"{url} decode error, aborting")
+        return -7, {}
-            else:
-                print(f"{url} decode error, aborting")
-            return 0, {}
-    else:
-        if logger:
-            logger.error(f"{url} undetected coding, aborting")
-        else:
-            print(f"{url} undetected coding, aborting")
-        return 0, {}
 
     result = extractor.extract(text)
     if not result:
-        if logger:
         logger.error(f"gne cannot extract {url}")
-        else:
-            print(f"gne cannot extract {url}")
         return 0, {}
 
-    if len(result['title']) < 5 or len(result['content']) < 24:
+    if len(result['title']) < 4 or len(result['content']) < 24:
-        if logger:
+        logger.info(f"{result} not valid")
-            logger.warning(f"{result} not valid")
-        else:
-            print(f"{result} not valid")
         return 0, {}
 
-    if result['title'].startswith('服务器错误') or result['title'].startswith('您访问的页面') or result['title'].startswith('403'):
+    if result['title'].startswith('服务器错误') or result['title'].startswith('您访问的页面') or result['title'].startswith('403')\
-        if logger:
+            or result['content'].startswith('This website uses cookies') or result['title'].startswith('出错了'):
         logger.warning(f"can not get {url} from the Internet")
-        else:
-            print(f"can not get {url} from the Internet")
         return -7, {}
 
-    date_str = re.findall(r"\d{4}-\d{2}-\d{2}", result['publish_time'])
+    date_str = extract_and_convert_dates(result['publish_time'])
     if date_str:
-        result['publish_time'] = date_str[0].replace("-", "")
+        result['publish_time'] = date_str
-    else:
-        date_str = re.findall(r"\d{4}\.\d{2}\.\d{2}", result['publish_time'])
-        if date_str:
-            result['publish_time'] = date_str[0].replace(".", "")
-        else:
-            date_str = re.findall(r"\d{4}\d{2}\d{2}", result['publish_time'])
-            if date_str:
-                result['publish_time'] = date_str[0]
     else:
         result['publish_time'] = datetime.strftime(datetime.today(), "%Y%m%d")
 
@@ -93,7 +57,7 @@ def simple_crawler(url: str | Path, logger=None) -> (int, dict):
             result['abstract'] = meta_description["content"]
         else:
             result['abstract'] = ''
-    except:
+    except Exception:
         result['abstract'] = ''
 
     result['url'] = str(url)

client/backend/work_process.py

@@ -59,13 +59,13 @@ class ServiceProcesser:
         # 定义扫描源列表，如果不指定就默认遍历scraper_map, 另外这里还要考虑指定的source不在scraper_map的情况，这时应该使用通用爬虫
         sources = sites if sites else list(scraper_map.keys())
         new_articles = []
-        with concurrent.futures.ThreadPoolExecutor(max_workers=3) as executor:
+        with concurrent.futures.ThreadPoolExecutor(max_workers=5) as executor:
             futures = []
             for site in sources:
                 if site in scraper_map:
-                    futures.append(executor.submit(scraper_map[site], expiration, existings, self.logger))
+                    futures.append(executor.submit(scraper_map[site], expiration, existings))
                 else:
-                    futures.append(executor.submit(general_scraper, site, expiration, existings, self.logger))
+                    futures.append(executor.submit(general_scraper, site, expiration, existings))
             concurrent.futures.wait(futures)
             for future in futures:
                 try:

client/config.ini

@@ -7,8 +7,3 @@ good_sample1 = 黑客组织Rhysida声称已入侵中国国有能源公司
 good_sample2 = 差不多一百万份包含未成年人数据（包括家庭地址和照片）的文件对互联网上的任何人都开放，对孩子构成威胁
 bad_sample = 黑客组织活动最近频发
 report_type = 网络安全情报
-
-[sites]
-site3 = https://www.hackread.com/
-site2 = http://sh.people.com.cn/
-site1 = https://www.xuexi.cn/

client/sites.txt

@@ -0,0 +1,5 @@
+https://www.hackread.com/
+http://sh.people.com.cn/
+https://www.xuexi.cn/
+https://www.defensenews.com/
+https://www.meritalk.com

client/version

@@ -1 +1 @@
-v0.2.0
+v0.2.1