pycoze 0.1.438__tar.gz → 0.1.440__tar.gz

{pycoze-0.1.438 → pycoze-0.1.440}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: pycoze
-Version: 0.1.438
+Version: 0.1.440
 Summary: Package for pycoze only!
 Author: Yuan Jie Xiong
 Author-email: aiqqqqqqq@qq.com

pycoze-0.1.440/pycoze/api/lib/web.py

@@ -0,0 +1,65 @@
+from pycoze import utils
+from bs4 import BeautifulSoup, Comment
+
+
+socket = utils.socket
+
+
+class WebCls:
+    def get_simplified_webpage(self, url: str) -> str:
+        return socket.post_and_recv_result(
+            "getSimplifiedWebpage", {"url": url}
+        )
+    
+    def get_simplified_html(self, html: str, selector=None) -> str:
+        soup = BeautifulSoup(html, 'html.parser')
+
+        # 如果指定了selector，则只提取该元素的内容
+        if selector:
+            element = soup.select_one(selector)
+            if element:
+                soup = BeautifulSoup(str(element), 'html.parser')
+
+        # 定义需要移除的标签
+        tags_to_remove = ['script', 'style', 'noscript', 'meta', 'link']
+        for tag in tags_to_remove:
+            for element in soup(tag):
+                element.decompose()
+
+        # 移除注释
+        for element in soup.find_all(string=lambda text: isinstance(text, Comment)):
+            element.extract()
+
+        # 定义需要保留的交互属性
+        INTERACTIVE_ATTRIBUTES = {
+            'a': ['href', 'onclick'],
+            'button': ['onclick'],
+            'img': ['src', 'onload'],
+            'form': ['action', 'onsubmit'],
+            'input': ['type', 'onclick', 'onchange'],
+            '*': ['onclick', 'onload', 'onchange', 'onsubmit', 'onmouseover']
+        }
+
+        # 遍历所有标签，保留交互属性并移除其他属性
+        for element in soup.find_all(True):
+            tag_name = element.name
+            allowed_attrs = INTERACTIVE_ATTRIBUTES.get(tag_name, []) + INTERACTIVE_ATTRIBUTES['*']
+            attrs = list(element.attrs.keys())
+            for attr in attrs:
+                if attr not in allowed_attrs:
+                    del element[attr]
+
+            # 如果是<img>标签，检查src是否为Base64
+            if tag_name == 'img' and 'src' in element.attrs and element['src'].startswith('data:'):
+                del element['src']
+
+            # 处理文本内容，超过1000字符则截取
+            if element.string and len(element.string) > 1000:
+                element.string = element.string[:1000] + '...'
+
+        # 移除标签之间的多余空白
+        for element in soup.find_all(True):
+            if not element.get_text(strip=True):
+                element.string = ''
+
+        return str(soup)

{pycoze-0.1.438 → pycoze-0.1.440}/pycoze/bot/lib.py

@@ -66,10 +66,8 @@ def get_system_prompt(abilities, bot_setting):
         cd_prompt = "When executing outside the working directory, include the CD command, such as cd /path/to/directory && ls."
     system = platform.system()
 
-    print("bot_setting", bot_setting)
-    print("abilities", abilities)
+
     abilities_str = "\n".join([function_to_string(a) for a in abilities])
-    print("abilities_str", abilities_str)
 
     context = {
         "prompt": bot_setting["prompt"],
@@ -96,7 +94,7 @@ def get_system_prompt(abilities, bot_setting):
         if key != "folder_context" and bot_setting["systemAbility"][key] == True:
             has_any_tool = True
             break
-
+            
     context["has_any_tool"] = has_any_tool
     system_prompt = template.render(context)
 

{pycoze-0.1.438 → pycoze-0.1.440}/pycoze.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: pycoze
-Version: 0.1.438
+Version: 0.1.440
 Summary: Package for pycoze only!
 Author: Yuan Jie Xiong
 Author-email: aiqqqqqqq@qq.com

{pycoze-0.1.438 → pycoze-0.1.440}/setup.py

@@ -2,7 +2,7 @@ from setuptools import setup, find_packages
 
 setup(
     name="pycoze",
-    version="0.1.438",
+    version="0.1.440",
     packages=find_packages(),
     install_requires=[],
     author="Yuan Jie Xiong",

pycoze-0.1.438/pycoze/api/lib/web.py

@@ -1,16 +0,0 @@
-from pycoze import utils
-
-
-socket = utils.socket
-
-
-class WebCls:
-    def get_simplified_webpage(self, url: str) -> str:
-        return socket.post_and_recv_result(
-            "getSimplifiedWebpage", {"url": url}
-        )
-    
-    def get_simplified_html(self, html: str) -> str:
-        return socket.post_and_recv_result(
-            "getSimplifiedHtml", {"html": html}
-        )