jarvis-ai-assistant 0.1.208__py3-none-any.whl → 0.1.210__py3-none-any.whl

jarvis/jarvis_platform/tongyi.py

@@ -5,9 +5,8 @@ import time
 import uuid
 from typing import Any, Dict, Generator, List, Tuple
 
-import requests  # type: ignore
-
 from jarvis.jarvis_platform.base import BasePlatform
+from jarvis.jarvis_utils import http
 from jarvis.jarvis_utils.output import OutputType, PrettyOutput
 from jarvis.jarvis_utils.utils import while_success
 
@@ -160,82 +159,93 @@ class TongyiPlatform(BasePlatform):
         }
 
         try:
-            response = while_success(
-                lambda: requests.post(url, headers=headers, json=payload, stream=True),
+            # 使用新的stream_post接口发送消息请求，获取流式响应
+            response_stream = while_success(
+                lambda: http.stream_post(url, headers=headers, json=payload),
                 sleep_time=5,
             )
-            if response.status_code != 200:
-                raise Exception(f"HTTP {response.status_code}: {response.text}")
+
             msg_id = ""
             session_id = ""
             thinking_content = ""
             text_content = ""
             in_thinking = False
-            for line in response.iter_lines():
-                if not line:
-                    continue
-                line_str = line.decode("utf-8")
-                if not line_str.startswith("data: "):
-                    continue
+            response_data = b""
+
+            # 处理流式响应
+            for chunk in response_stream:
+                response_data += chunk
 
+                # 尝试解析SSE格式的数据
                 try:
-                    data = json.loads(line_str[6:])
-                    # 记录消息ID和会话ID
-                    if "msgId" in data:
-                        msg_id = data["msgId"]
-                    if "sessionId" in data:
-                        session_id = data["sessionId"]
-
-                    if "contents" in data and len(data["contents"]) > 0:
-                        for content in data["contents"]:
-                            if content.get("contentType") == "think":
-                                if not in_thinking:
-                                    yield "<think>\n\n"
-                                    in_thinking = True
-                                if content.get("incremental"):
-                                    tmp_content = json.loads(content.get("content"))[
-                                        "content"
-                                    ]
-                                    thinking_content += tmp_content
-                                    yield tmp_content
-                                else:
-                                    tmp_content = json.loads(content.get("content"))[
-                                        "content"
-                                    ]
-                                    if len(thinking_content) < len(tmp_content):
-                                        yield tmp_content[len(thinking_content) :]
-                                        thinking_content = tmp_content
-                                    else:
-                                        # thinking_content = "aaa</thi"
-                                        # tmp_content = "aaa"
-                                        # 应该yield nk>
-                                        # print("\n")
-                                        # print(len(thinking_content))
-                                        # print(len(tmp_content))
-                                        # print("--------------------------------")
-                                        # print(thinking_content)
-                                        # print("--------------------------------")
-                                        # print(tmp_content)
-                                        # print("--------------------------------")
-                                        yield "\r\n</think>\n"[
-                                            len(thinking_content) - len(tmp_content) :
-                                        ]
-                                        thinking_content = tmp_content
-                                    in_thinking = False
-                            elif content.get("contentType") == "text":
-                                if in_thinking:
-                                    continue
-                                if content.get("incremental"):
-                                    tmp_content = content.get("content")
-                                    text_content += tmp_content
-                                    yield tmp_content
-                                else:
-                                    tmp_content = content.get("content")
-                                    if len(text_content) < len(tmp_content):
-                                        yield tmp_content[len(text_content) :]
-                                        text_content = tmp_content
-
-                except json.JSONDecodeError:
+                    # 查找完整的数据行
+                    lines = response_data.decode("utf-8").split("\n")
+                    response_data = b""  # 重置缓冲区
+
+                    for line in lines:
+                        if not line.strip():
+                            continue
+
+                        # SSE格式的行通常以"data: "开头
+                        if line.startswith("data: "):
+                            try:
+                                data = json.loads(line[6:])
+                                # 记录消息ID和会话ID
+                                if "msgId" in data:
+                                    msg_id = data["msgId"]
+                                if "sessionId" in data:
+                                    session_id = data["sessionId"]
+
+                                if "contents" in data and len(data["contents"]) > 0:
+                                    for content in data["contents"]:
+                                        if content.get("contentType") == "think":
+                                            if not in_thinking:
+                                                yield "<think>\n\n"
+                                                in_thinking = True
+                                            if content.get("incremental"):
+                                                tmp_content = json.loads(
+                                                    content.get("content")
+                                                )["content"]
+                                                thinking_content += tmp_content
+                                                yield tmp_content
+                                            else:
+                                                tmp_content = json.loads(
+                                                    content.get("content")
+                                                )["content"]
+                                                if len(thinking_content) < len(
+                                                    tmp_content
+                                                ):
+                                                    yield tmp_content[
+                                                        len(thinking_content) :
+                                                    ]
+                                                    thinking_content = tmp_content
+                                                else:
+                                                    yield "\r\n</think>\n"[
+                                                        len(thinking_content)
+                                                        - len(tmp_content) :
+                                                    ]
+                                                    thinking_content = tmp_content
+                                                in_thinking = False
+                                        elif content.get("contentType") == "text":
+                                            if in_thinking:
+                                                continue
+                                            if content.get("incremental"):
+                                                tmp_content = content.get("content")
+                                                text_content += tmp_content
+                                                yield tmp_content
+                                            else:
+                                                tmp_content = content.get("content")
+                                                if len(text_content) < len(tmp_content):
+                                                    yield tmp_content[
+                                                        len(text_content) :
+                                                    ]
+                                                    text_content = tmp_content
+
+                            except json.JSONDecodeError:
+                                continue
+
+                except UnicodeDecodeError:
+                    # 如果解码失败，继续累积数据
                     continue
 
             self.msg_id = msg_id
@@ -258,7 +268,7 @@ class TongyiPlatform(BasePlatform):
 
         try:
             response = while_success(
-                lambda: requests.post(url, headers=headers, json=payload), sleep_time=5
+                lambda: http.post(url, headers=headers, json=payload), sleep_time=5
             )
             if response.status_code != 200:
                 raise Exception(f"HTTP {response.status_code}: {response.text}")
@@ -314,7 +324,7 @@ class TongyiPlatform(BasePlatform):
                     print(f"📤 正在上传文件: {file_name}")
 
                     # Upload file
-                    response = requests.post(
+                    response = http.post(
                         upload_token["host"], data=form_data, files=files
                     )
 
@@ -349,7 +359,7 @@ class TongyiPlatform(BasePlatform):
                         "dir": upload_token["dir"],
                     }
 
-                    response = requests.post(url, headers=headers, json=payload)
+                    response = http.post(url, headers=headers, json=payload)
                     if response.status_code != 200:
                         print(f"❌ 获取下载链接失败: HTTP {response.status_code}")
                         return False
@@ -381,7 +391,7 @@ class TongyiPlatform(BasePlatform):
                             "fileSize": os.path.getsize(file_path),
                         }
 
-                        add_response = requests.post(
+                        add_response = http.post(
                             add_url, headers=headers, json=add_payload
                         )
                         if add_response.status_code != 200:
@@ -464,7 +474,7 @@ class TongyiPlatform(BasePlatform):
 
         try:
             response = while_success(
-                lambda: requests.post(url, headers=headers, json=payload), sleep_time=5
+                lambda: http.post(url, headers=headers, json=payload), sleep_time=5
             )
             if response.status_code != 200:
                 PrettyOutput.print(

jarvis/jarvis_platform/yuanbao.py

@@ -7,10 +7,10 @@ import time
 import urllib.parse
 from typing import Dict, Generator, List, Tuple
 
-import requests  # type: ignore
 from PIL import Image  # type: ignore
 
 from jarvis.jarvis_platform.base import BasePlatform
+from jarvis.jarvis_utils import http
 from jarvis.jarvis_utils.output import OutputType, PrettyOutput
 from jarvis.jarvis_utils.utils import while_success
 
@@ -37,7 +37,7 @@ class YuanbaoPlatform(BasePlatform):
         self.conversation_id = ""  # 会话ID，用于标识当前对话
         # 从环境变量中获取必要参数
         self.cookies = os.getenv("YUANBAO_COOKIES")  # 认证cookies
-        self.agent_id = os.getenv("YUANBAO_AGENT_ID")  # 代理ID
+        self.agent_id = "naQivTmsDa"
 
         if not self.cookies:
             PrettyOutput.print("YUANBAO_COOKIES 未设置", OutputType.WARNING)
@@ -95,7 +95,7 @@ class YuanbaoPlatform(BasePlatform):
 
         try:
             response = while_success(
-                lambda: requests.post(url, headers=headers, data=payload, timeout=600),
+                lambda: http.post(url, headers=headers, data=payload),
                 sleep_time=5,
             )
             response_json = response.json()
@@ -254,7 +254,7 @@ class YuanbaoPlatform(BasePlatform):
 
         try:
             response = while_success(
-                lambda: requests.post(url, headers=headers, json=payload, timeout=600),
+                lambda: http.post(url, headers=headers, json=payload),
                 sleep_time=5,
             )
 
@@ -331,7 +331,7 @@ class YuanbaoPlatform(BasePlatform):
             )
 
             # Upload the file
-            response = requests.put(url, headers=headers, data=file_content)
+            response = http.put(url, headers=headers, data=file_content)
 
             if response.status_code not in [200, 204]:
                 PrettyOutput.print(
@@ -468,60 +468,66 @@ class YuanbaoPlatform(BasePlatform):
             payload["displayPrompt"] = payload["prompt"]
 
         try:
-            # 发送消息请求，获取流式响应
-            response = while_success(
-                lambda: requests.post(
-                    url, headers=headers, json=payload, stream=True, timeout=600
-                ),
+            # 使用新的stream_post接口发送消息请求，获取流式响应
+            response_stream = while_success(
+                lambda: http.stream_post(url, headers=headers, json=payload),
                 sleep_time=5,
             )
 
-            # 检查响应状态
-            if response.status_code != 200:
-                error_msg = f"发送消息失败，状态码: {response.status_code}"
-                if hasattr(response, "text"):
-                    error_msg += f", 响应: {response.text}"
-                raise Exception(error_msg)
-
             in_thinking = False
-
-            # 处理SSE流响应
-            for line in response.iter_lines():
-                if not line:
+            response_data = b""
+
+            # 处理流式响应
+            for chunk in response_stream:
+                response_data += chunk
+
+                # 尝试解析SSE格式的数据
+                try:
+                    # 查找完整的数据行
+                    lines = response_data.decode("utf-8").split("\n")
+                    response_data = b""  # 重置缓冲区
+
+                    for line in lines:
+                        if not line.strip():
+                            continue
+
+                        # SSE格式的行通常以"data: "开头
+                        if line.startswith("data: "):
+                            try:
+                                data_str = line[6:]  # 移除"data: "前缀
+
+                                # 检查结束标志
+                                if data_str == "[DONE]":
+                                    self.first_chat = False
+                                    return None
+
+                                data = json.loads(data_str)
+
+                                # 处理文本类型的消息
+                                if data.get("type") == "text":
+                                    if in_thinking:
+                                        yield "</think>\n"
+                                        in_thinking = False
+                                    msg = data.get("msg", "")
+                                    if msg:
+                                        yield msg
+
+                                # 处理思考中的消息
+                                elif data.get("type") == "think":
+                                    if not in_thinking:
+                                        yield "<think>\n"
+                                        in_thinking = True
+                                    think_content = data.get("content", "")
+                                    if think_content:
+                                        yield think_content
+
+                            except json.JSONDecodeError:
+                                pass
+
+                except UnicodeDecodeError:
+                    # 如果解码失败，继续累积数据
                     continue
 
-                line_str = line.decode("utf-8")
-
-                # SSE格式的行通常以"data: "开头
-                if line_str.startswith("data: "):
-                    try:
-                        data_str = line_str[6:]  # 移除"data: "前缀
-                        data = json.loads(data_str)
-
-                        # 处理文本类型的消息
-                        if data.get("type") == "text":
-                            if in_thinking:
-                                yield "</think>\n"
-                                in_thinking = False
-                            msg = data.get("msg", "")
-                            if msg:
-                                yield msg
-
-                        # 处理思考中的消息
-                        elif data.get("type") == "think":
-                            if not in_thinking:
-                                yield "<think>\n"
-                                in_thinking = True
-                            think_content = data.get("content", "")
-                            if think_content:
-                                yield think_content
-
-                    except json.JSONDecodeError:
-                        pass
-
-                # 检测结束标志
-                elif line_str == "data: [DONE]":
-                    return None
             self.first_chat = False
             return None
 
@@ -547,7 +553,7 @@ class YuanbaoPlatform(BasePlatform):
 
         try:
             response = while_success(
-                lambda: requests.post(url, headers=headers, json=payload, timeout=600),
+                lambda: http.post(url, headers=headers, json=payload),
                 sleep_time=5,
             )
 

jarvis/jarvis_tools/ask_user.py

@@ -34,7 +34,6 @@ class AskUserTool:
 
             # 获取agent对象并重置工具调用计数
             agent = args["agent"]
-            agent.reset_tool_call_count()
 
             # 显示问题给用户
             PrettyOutput.print(f"问题: {question}", OutputType.SYSTEM)

jarvis/jarvis_tools/file_analyzer.py

@@ -45,9 +45,6 @@ class FileAnalyzerTool:
             file_paths = args["file_paths"]
             prompt = args["prompt"]
 
-            agent = args["agent"]
-            agent.reset_tool_call_count()
-
             # 验证文件路径
             valid_files = []
             for file_path in file_paths:

jarvis/jarvis_utils/config.py

@@ -3,7 +3,7 @@ import os
 from functools import lru_cache
 from typing import Any, Dict, List
 
-import yaml
+import yaml  # type: ignore
 
 from jarvis.jarvis_utils.builtin_replace_map import BUILTIN_REPLACE_MAP
 
@@ -96,16 +96,6 @@ def get_max_input_token_count() -> int:
     return int(GLOBAL_CONFIG_DATA.get("JARVIS_MAX_INPUT_TOKEN_COUNT", "32000"))
 
 
-def is_auto_complete() -> bool:
-    """
-    检查是否启用了自动补全功能。
-
-    返回：
-        bool: 如果启用了自动补全则返回True，默认为False
-    """
-    return GLOBAL_CONFIG_DATA.get("JARVIS_AUTO_COMPLETE", False) == True
-
-
 def get_shell_name() -> str:
     """
     获取系统shell名称。
@@ -119,10 +109,6 @@ def get_shell_name() -> str:
     3. 最后从环境变量SHELL获取
     4. 如果都未配置，则默认返回bash
     """
-    shell_name = GLOBAL_CONFIG_DATA.get("JARVIS_SHELL")
-    if shell_name:
-        return shell_name.lower()
-    
     shell_path = GLOBAL_CONFIG_DATA.get("SHELL", os.getenv("SHELL", "/bin/bash"))
     return os.path.basename(shell_path).lower()
 
@@ -191,16 +177,6 @@ def is_confirm_before_apply_patch() -> bool:
     return GLOBAL_CONFIG_DATA.get("JARVIS_CONFIRM_BEFORE_APPLY_PATCH", False) == True
 
 
-def get_max_tool_call_count() -> int:
-    """
-    获取最大工具调用次数。
-
-    返回：
-        int: 最大连续工具调用次数，默认为20
-    """
-    return int(GLOBAL_CONFIG_DATA.get("JARVIS_MAX_TOOL_CALL_COUNT", "20"))
-
-
 def get_data_dir() -> str:
     """
     获取Jarvis数据存储目录路径。
@@ -209,20 +185,9 @@ def get_data_dir() -> str:
         str: 数据目录路径，优先从JARVIS_DATA_PATH环境变量获取，
              如果未设置或为空，则使用~/.jarvis作为默认值
     """
-    data_path = GLOBAL_CONFIG_DATA.get("JARVIS_DATA_PATH", "").strip()
-    if not data_path:
-        return os.path.expanduser("~/.jarvis")
-    return data_path
-
-
-def get_auto_update() -> bool:
-    """
-    获取是否自动更新git仓库。
-
-    返回：
-        bool: 如果需要自动更新则返回True，默认为True
-    """
-    return GLOBAL_CONFIG_DATA.get("JARVIS_AUTO_UPDATE", True) == True
+    return os.path.expanduser(
+        GLOBAL_CONFIG_DATA.get("JARVIS_DATA_PATH", "~/.jarvis").strip()
+    )
 
 
 def get_max_big_content_size() -> int:
@@ -275,16 +240,6 @@ def is_print_prompt() -> bool:
     return GLOBAL_CONFIG_DATA.get("JARVIS_PRINT_PROMPT", False) == True
 
 
-def get_history_count() -> int:
-    """
-    获取是否启用历史记录功能。
-
-    返回：
-        bool: 如果启用历史记录则返回True，默认为False
-    """
-    return GLOBAL_CONFIG_DATA.get("JARVIS_USE_HISTORY_COUNT", 0)
-
-
 def get_mcp_config() -> List[Dict[str, Any]]:
     """
     获取MCP配置列表。

jarvis/jarvis_utils/embedding.py

@@ -1,17 +1,11 @@
 # -*- coding: utf-8 -*-
-import functools
-import os
-from typing import Any, List
+from typing import List
 
-from jarvis.jarvis_utils.config import get_data_dir
 from jarvis.jarvis_utils.output import OutputType, PrettyOutput
 
-# 全局缓存，避免重复加载模型
-_global_tokenizers = {}
-
 
 def get_context_token_count(text: str) -> int:
-    """使用分词器获取文本的token数量。
+    """使用tiktoken获取文本的token数量。
 
     参数：
         text: 要计算token的输入文本
@@ -20,16 +14,10 @@ def get_context_token_count(text: str) -> int:
         int: 文本中的token数量
     """
     try:
-        from transformers import AutoTokenizer  # type: ignore
-        tokenizer : AutoTokenizer = load_tokenizer()
-        # 分批处理长文本，确保不超过模型最大长度
-        total_tokens = 0
-        chunk_size = 100  # 每次处理100个字符，避免超过模型最大长度（考虑到中文字符可能被编码成多个token）
-        for i in range(0, len(text), chunk_size):
-            chunk = text[i : i + chunk_size]
-            tokens = tokenizer.encode(chunk)  # type: ignore
-            total_tokens += len(tokens)
-        return total_tokens
+        import tiktoken
+
+        encoding = tiktoken.get_encoding("cl100k_base")
+        return len(encoding.encode(text))
     except Exception as e:
         PrettyOutput.print(f"计算token失败: {str(e)}", OutputType.WARNING)
         return len(text) // 4  # 每个token大约4个字符的粗略估计
@@ -84,36 +72,3 @@ def split_text_into_chunks(
         PrettyOutput.print(f"文本分割失败: {str(e)}", OutputType.WARNING)
         # 发生错误时回退到简单的字符分割
         return [text[i : i + max_length] for i in range(0, len(text), max_length)]
-
-
-@functools.lru_cache(maxsize=1)
-def load_tokenizer() -> Any:
-    """
-    加载用于文本处理的分词器，使用缓存避免重复加载。
-
-    返回：
-        AutoTokenizer: 加载的分词器
-    """
-
-    from transformers import AutoTokenizer  # type: ignore
-
-    model_name = "gpt2"
-    cache_dir = os.path.join(get_data_dir(), "huggingface", "hub")
-
-    # 检查全局缓存
-    if model_name in _global_tokenizers:
-        return _global_tokenizers[model_name]
-
-    try:
-        tokenizer = AutoTokenizer.from_pretrained(
-            model_name, cache_dir=cache_dir, local_files_only=True
-        )
-    except Exception:
-        tokenizer = AutoTokenizer.from_pretrained(
-            model_name, cache_dir=cache_dir, local_files_only=False
-        )
-
-    # 保存到全局缓存
-    _global_tokenizers[model_name] = tokenizer
-
-    return tokenizer  # type: ignore