pycoze 0.1.439__py3-none-any.whl → 0.1.440__py3-none-any.whl

pycoze/api/lib/web.py

@@ -1,4 +1,5 @@
 from pycoze import utils
+from bs4 import BeautifulSoup, Comment
 
 
 socket = utils.socket
@@ -10,7 +11,55 @@ class WebCls:
             "getSimplifiedWebpage", {"url": url}
         )
     
-    def get_simplified_html(self, html: str) -> str:
-        return socket.post_and_recv_result(
-            "getSimplifiedHtml", {"html": html}
-        )
+    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/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-0.1.439.dist-info → pycoze-0.1.440.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: pycoze
-Version: 0.1.439
+Version: 0.1.440
 Summary: Package for pycoze only!
 Author: Yuan Jie Xiong
 Author-email: aiqqqqqqq@qq.com

{pycoze-0.1.439.dist-info → pycoze-0.1.440.dist-info}/RECORD

@@ -7,12 +7,12 @@ pycoze/api/__init__.py,sha256=TLKvaZlRzTTt0KiXijLjj9b_iCr7fU1siwsXqyd74b8,375
 pycoze/api/lib/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 pycoze/api/lib/tab.py,sha256=DWO8ElI-VOODtIxqUFWaDB8VRrrFYAZRWivuIeD1wG0,2619
 pycoze/api/lib/view.py,sha256=_PIpTfeuTPPlMDKshMGsqFQYMq7ZiO4Hg5XwHwDoU60,7357
-pycoze/api/lib/web.py,sha256=gR2Ffaj3Pk4NfnE314UTLfZm7rW_rizZxWVDcZMeFgs,382
+pycoze/api/lib/web.py,sha256=l56koyd7RUzK-4mQE_5W-POyqSD79mvwusIN0rxND-Q,2347
 pycoze/api/lib/window.py,sha256=dkzWfLwn5pE_L0DfQ38K8nx9tQyT5KO-GYyXi0rytFc,2073
 pycoze/bot/__init__.py,sha256=rL3Q-ycczRpSFfKn84fg3QBl5k22WpyeIU5qOEjEby8,79
 pycoze/bot/chat.py,sha256=qEuMxH0cVFU9QSU36FrOsjhRAxtsvOv7CQtuXvM3F6Y,6446
 pycoze/bot/chat_base.py,sha256=uwAVhhvpKHQytJYMsIOHsTThkr45xRehbXF0-kDbvjM,12614
-pycoze/bot/lib.py,sha256=-9zeP2GMxUtyxBbBUwU3VEybalzZnNZGOROImWftJ7E,7311
+pycoze/bot/lib.py,sha256=_bQ52mKsWgFGAogFHnmRBJbvK_tPOwsAJ8NqJNMR5K4,7210
 pycoze/bot/message.py,sha256=udnIi-h4QgGzkbr_5VcAsVGjoLp9wXJSfBCeuOz7_Bk,802
 pycoze/bot/prompt.md,sha256=t7NQdiiNe-jCDVfeVbvTPfq5WK5nF8CxFUQUFMyXJlo,13880
 pycoze/bot/tools.py,sha256=uFvGIQ4VzkiH-TM9U9RenWfYVYyeUv_bP-094hB9rTc,10367
@@ -31,8 +31,8 @@ pycoze/utils/arg.py,sha256=jop1tBfe5hYkHW1NSpCeaZBEznkgguBscj_7M2dWfrs,503
 pycoze/utils/env.py,sha256=5pWlXfM1F5ZU9hhv1rHlDEanjEW5wf0nbyez9bNRqqA,559
 pycoze/utils/socket.py,sha256=bZbFFRH4mfThzRqt55BAAGQ6eICx_ja4x8UGGrUdAm8,2428
 pycoze/utils/text_or_file.py,sha256=gpxZVWt2DW6YiEg_MnMuwg36VNf3TX383QD_1oZNB0Y,551
-pycoze-0.1.439.dist-info/LICENSE,sha256=QStd_Qsd0-kAam_-sOesCIp_uKrGWeoKwt9M49NVkNU,1090
-pycoze-0.1.439.dist-info/METADATA,sha256=Dt-scKTwLVWtUToYmBJDaL-i1oJEl8U2vI1o-Rt7Mr0,854
-pycoze-0.1.439.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-pycoze-0.1.439.dist-info/top_level.txt,sha256=76dPeDhKvOCleL3ZC5gl1-y4vdS1tT_U1hxWVAn7sFo,7
-pycoze-0.1.439.dist-info/RECORD,,
+pycoze-0.1.440.dist-info/LICENSE,sha256=QStd_Qsd0-kAam_-sOesCIp_uKrGWeoKwt9M49NVkNU,1090
+pycoze-0.1.440.dist-info/METADATA,sha256=c-1yayy-SaotLhwt9UK_dm7XGCa9uEhnaSeuGA4Gq10,854
+pycoze-0.1.440.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+pycoze-0.1.440.dist-info/top_level.txt,sha256=76dPeDhKvOCleL3ZC5gl1-y4vdS1tT_U1hxWVAn7sFo,7
+pycoze-0.1.440.dist-info/RECORD,,